Quality measures are increasingly used to guide clinicians’ treatment decisions and benchmark hospitals against national averages and peer institutions.1,2 Implementation and high performance on these measures are required for hospital and program accreditation and pay-for-performance reimbursement.3,4,5,6,7 They also allow for standardization of care and the ability to evaluate outcomes across providers and institutions.8 Historically, cancer quality measures have focused on the active provision of care and adherence to guidelines for recommended therapies (i.e., delivering adjuvant chemotherapy or offering genetic testing when indicated). More recently, there has been increasing recognition of cancer overtreatment and the financial toxicities associated with low-value cancer care in the USA.5,6,7,8 In response, organizations such as the Center for Medicare & Medicaid Services (CMS) are now prioritizing the identification of quality measures that will reduce low-value care through their value-based programs centered on lowering costs and unnecessary care for patients.9

There is clear importance for improving patient care in the understanding and appropriate implementation of quality measure in healthcare. Whether this prioritization has resulted in identification of more quality measures aimed at avoiding low-value care is still unknown. Previous studies evaluating cancer quality measures only used a subset of measures or were not focused on measures promoting the avoidance of low-value care. Cancer quality measures have previously been evaluated to identify the most impactful measures10,11 and assess their reliability12, implementation13,14, practicality15, or cost-effectiveness.16 One study described the utility of quality measures aimed at avoiding low-value care but only cited a few measures.17

Our objective was to evaluate to what degree avoiding overtreatment or low-value care is viewed as a quality indicator by major quality organizations and accrediting bodies. Through review of leading quality measure organizations, we aimed to identify, classify, and review cancer-specific quality measures centered on the avoidance of low-value care. Secondarily, we aimed to identify gaps in current measures that may focus efforts for development of future quality measures.

Methods

We conducted a review of published cancer-specific quality measures from August 2019 to February 2020. We obtained an initial list of measures from a recently published comprehensive review of cancer quality measures,1 which included 300 measures from the National Quality Forum (NQF) or the National Quality Measures Clearinghouse (NQMC). The NQF and NQMC quality measure databases were reviewed to ensure this list of quality measures was accurate.18,19 To expand our review, we also reviewed four other quality measure programs to find additional cancer-specific quality measures: the CMS Merit-based Incentive Payment System (MIPS), the National Committee for Quality Assurance’s (NCQA) Healthcare Effectiveness Data and Information Set (HEDIS), the American Society of Clinical Oncology’s (ASCO) Quality Oncology Practice Initiative (QOPI), and the Commission on Cancer (CoC) quality measures.20,21,22,23 We included an additional 35 measures following this process. We felt this was an expansive search since the NQF-endorsed quality measures make up approximately 50% of quality measures used by federal healthcare programs and more than 30% of private payer programs.24 The remaining organizations were commonly referenced quality measure programs.11,25,26

Two members of the study team (B.L.E. and A.K.M.) reviewed and classified the quality measures. Classification categories included non-disease site-specific (e.g., general oncologic care, symptoms and toxicities, end-of-life care, and palliative care) and disease site-specific (e.g., breast cancer, colorectal cancer, lung cancer, prostate cancer, etc.) measures. The grouping of non-disease site-specific quality measures follows ASCO’s categorization of nonspecific quality measures.23 End-of-life care quality measures were specific to the last 30 days of life. Measures were additionally classified on the basis of the phase of care they targeted, which included: screening, diagnostic testing and staging, treatment, surveillance, and clinical outcomes.

Due to many cancer types having a low number of quality measures, a threshold for cancer type inclusion was set at ten unique quality measures. This resulted in the inclusion and review of four cancer types: breast, colorectal, lung, and prostate cancer. Lung cancer measures included those specific to non-small cell (NSCLC) and small cell lung cancer (SCLC) as well as general lung cancer treatment.

Starting with a previously described definition of low-value care,27 measures promoting avoidance of low-value care were defined as having significant evidence validating their effectiveness at reducing unnecessary care and decreasing healthcare resource utilization (examples included in Table 1). Quality measures were independently reviewed, and any disagreements in classification were discussed until a consensus was reached. Descriptive statistics were used in analysis. The University of Michigan Institutional Review Board deemed this project not regulated (not human subjects research).

Table 1 The number and percentages of cancer-specific quality measures promoting the avoidance of low-value care. Further categorized into the phase of care the measures target, with examples
Full size table

Results

A total of 313 quality measures met inclusion criteria (Fig. 1). Overall, 18% (n = 55) of the measures promoted the avoidance of low-value care, 73% (n = 230) promoted implementation of some aspect of care, and the remaining 9% (n = 28) were classified as “other.” The largest percentage of non-disease site quality measures involved general cancer care [n = 117 (37%)], followed by end-of-life care [n = 26 (8%)], palliative care [n = 18 (6%)], and symptoms and toxicities [n = 5 (2%)]. Breast cancer had the most disease site-specific quality measures [n = 65 (21%)], followed by colon [n = 40 (13%)], lung [n = 29 (9%)], and prostate [n = 13 (4%)] cancer. A list of quality measures promoting the avoidance of low-value care is provided in Appendix 1.

Fig. 1
figure 1

Flow diagram showing the phases of quality measure identification and selection

Full size image

For non-disease site-specific quality measures, end-of-life care had the most quality measures promoting the avoidance of low-value care (n = 13), followed by general care (n = 4), palliative care (n = 2), and symptoms and toxicities (n = 2). Breast cancer had the most disease site-specific low-value care avoiding quality measures (n = 12), superseding lung (n = 9), colon (n = 8), and prostate (n = 5) cancer.

Of the 55 quality measures promoting the avoidance of low-value care, 5 (9%) measures focused on screening, 7 (13%) on diagnostic testing and staging, 19 (34%) on treatment, 6 (11%) on surveillance, and 18 (33%) on clinical outcomes. Results are summarized in Table 1 and Figs. 2, 3, 4, 5.

Fig. 2
figure 2

The percentage of quality measures promoting the avoidance of low-value care, implementing care, or other for each category of non-disease site-specific cancer quality measures

Full size image
Fig. 3
figure 3

The percentage of quality measures promoting the avoidance of low-value care, implementing care, or other for each category of disease site-specific cancer quality measures

Full size image
Fig. 4
figure 4

The number of quality measures promoting the avoidance of low-value cancer care for each non-disease site-specific category, separated by the phase of care the measures target

Full size image
Fig. 5
figure 5

The number of quality measures promoting the avoidance of low-value cancer care for each disease site-specific category, separated by the phase of care the measures target

Full size image

Current gaps in quality measures promoting the avoidance of low-value care are presented in Table 2. There was no category where a representative quality measure was found for all phases of care, and breast cancer was the only category where four of the five phases were represented. For each category, a treatment-specific quality measure was found. Non-disease site-specific categories only had quality measures focused on treatment and/or clinical outcome phases of care. Diagnostic testing and staging-specific quality measures were found in the least number of categories, appearing in only breast and prostate cancer.

Table 2 The different phases of care with a representative quality measure promoting the avoidance of low-value care for each quality measure category
Full size table

Non-Disease Site-Specific Cancer Quality Measures

General

Of 117 general cancer quality measures, 4 promoted the avoidance of low-value care. The was only one treatment-specific quality measure, which focused on reducing administration of granulocyte colony-stimulating factor (GCSF) to patients receiving chemotherapy for metastatic cancer. GCSF is currently overused as prophylaxis against febrile neutropenia in patients receiving chemotherapy.28,29,30,31 For the cancers included in this study, metastatic breast, lung, and colon cancer treatment protocols do not endorse the routine use of GCSF.29,32,33 For potentially curable metastatic prostate cancer, GCSF is included in some guidelines, but its uses in these scenarios are limited.34 Three of the measures targeted clinical outcomes. One measured the number of emergency room visits per chemotherapy patient per year, the other two measured all-cause readmission rates for patients being treated for cancer.

Symptoms and Toxicities

Of five quality measures related to symptoms and toxicities, two promoted the avoidance of low-value care. One measure targeted treatment, specifically avoiding overuse of antiemetic therapy for low-risk antineoplastic agents. Many cancer treatment protocols include antiemetic regimens, but overuse of antinausea medications is common in low-risk chemotherapy regimens.35,36 The remaining measure centered on clinical outcomes, quantifying the number of cancer patients hospitalized for treatment-related symptoms.

Palliative Care

Of 18 palliative care quality measures, 2 promoted the avoidance of low-value care, both measures focused on low-value treatment. One aimed to reduce administration of chemotherapy to patients with metastatic cancer and an undocumented Eastern Cooperative Oncology Group (ECOG) performance status or a score of 3 or 4. The other measure intended to reduce the number of cancer patients with bone metastases receiving multiple-fraction radiation therapy, which is associated with higher complications rates than and equivalent outcomes to single-fraction radiotherapy.37

End-of-Life Care

Of 26 end-of-life care quality measures, 13 promoted the avoidance of low-value care. Two measures were treatment-specific, with both measuring the number of patients receiving chemotherapy near the end of life, one at 14 days and the other at 30 days before death. The other 12 measured clinical outcomes. Five quality measures emphasized early enrollment of cancer patients with limited life expectancy into hospice care. Hospice enrollment was measured by determining the average number of days a patient was enrolled in hospice care, the proportion of patients enrolled into hospice care at a predetermined number of days before death, or the proportion of patients not admitted to hospice care before death. Four measures intended to reduce the number of cancer patients going to the emergency room, being admitted to the hospital, or dying in an acute care setting in the last 30 days of life. Three measures assessed chemotherapy administration in the final 30 days of life to reduce its use during this time. Lastly, one measure evaluated the overall costs for a cancer patient in the last 30 days of life.

Disease Site-Specific Cancer Quality Measures

Breast Cancer

Of 65 breast cancer quality measures, 12 promoted the avoidance of low-value care. One of 12 measures targeted screening, specifically referring to avoiding the inappropriate use of “probably benign” [otherwise known as Breast Imaging Reporting and Data System-3 (BI-RADS-3)] in the assessment of breast imaging as this term has been shown to result in unnecessary referrals to breast cancer practitioners and to induce patient anxiety.38 Diagnostic testing and staging-related measures accounted for 33% (n = 4) of the measures. Staging considerations consists of avoiding positron emission tomography (PET), computed tomography (CT), or radionuclide bone scans within 60 days after diagnosis of early stage (stage I, IIA, or IIB) breast cancer. Diagnostic testing measures encouraged avoiding excisional biopsies in favor of needle biopsies, and measured the percentage of clinically node negative patients with stage T1–T2 disease who received a sentinel lymph node biopsy (SLNB) with the goal of reducing complete axillary lymph node dissection (ALND). SLNB has been shown to have equivalent overall survival and recurrence rates with less morbidity when compared with ALND.39,40 Treatment-related measures (n = 3) included not administering GCSF to patients who received chemotherapy for metastatic breast cancer, sparing patients with negative or undocumented human epidermal growth factor receptor 2 (HER2) status from trastuzumab treatment, or sparing patients with negative or undocumented estrogen/progesterone receptor status treatment from tamoxifen/aromatase inhibitor therapy. Surveillance measures (n = 4) focused on minimizing the use of PET, CT, or radionuclide bone scans and serum tumor markers within a year following diagnosis of breast cancer in patients who received treatment with curative intent. Both of these practices have been shown to have no clinical benefit and are not recommended in asymptomatic patients who are followed with the recommended frequency of examinations and mammography.41

Colon Cancer

Of 40 colon cancer quality measures, 8 (20%) promoted the avoidance of low-value care. Three measures involved low-value screening practices. Screening quality measures encouraged consideration of life expectancy and risks before screening patients aged 76–85 years with colonoscopy, withdrawing patients aged ≥ 86 years from screening, and ensuring patients aged 50–75 years who had a colonoscopy without biopsy or polypectomy have a recommended follow-up interval of at least 10 years before repeat colonoscopy. Treatment-specific measures (n = 3) involved not administering GCSF to patients who received chemotherapy for metastatic colon cancer and sparing patients with metastatic or nonmetastatic disease and positive KRAS or NRAS mutations treatment with anti-epidermal growth factor receptor monoclonal antibodies (Anti-EGFR MoAb). The one surveillance quality measure promoted the avoidance of PET or PET–CT scans in asymptomatic patients treated with curative intent. The one clinical outcomes-based measure assessed the facility rate of risk-standardized, all-cause, unplanned hospital visits within 7 days of an outpatient colonoscopy motivated by research suggesting clinicians performing colonoscopy commonly underestimate their complication rates.42,43

Lung Cancer (NSCLC, SCLC, and General)

Of 29 lung cancer quality measures, 9 promoted the avoidance of low-value care. Overall, seven addressed NSCLC, one focused on SCLC, and one targeted general lung cancer care. Six measures focused on treatment. As with measures for breast and colon cancer, one measure focused on refraining from administering GCSF to patients who received chemotherapy for metastatic NSCLC. Other NSCLC treatment-specific measures included avoiding adjuvant chemotherapy for patients with stage IA disease; avoiding adjuvant radiation for patients with stage IB or II cancer; not giving bevacizumab to patients with the American Joint Committee on Cancer (AJCC) stage IV or distant metastatic disease and squamous histology; and not prescribing EGFR tyrosine kinase inhibitor or ALK inhibitor therapy in patients with stage IV disease with negative or undocumented EGFR or ALK mutations. The one measure regarding treating SCLC focused on avoiding overtreatment with platinum-based chemotherapy. While, traditionally, recommendations for the number of cycles of platinum-based chemotherapy in SCLC have ranged from 4 to 6, recent evidence has shown that four cycles may be ideal at balancing effectiveness and risk.44 The one measure that targeted surveillance was to avoid PET or PET–CT within 12 months after treatment with curative intent for patients with stage I or II NSCLC. Two clinical outcomes-related measures were identified, which emphasized the reduction of readmissions and overall complications after elective lobectomy, respectively.

Prostate Cancer

Of 13 prostate cancer quality measures, 5 promoted the avoidance of low-value care. The one screening quality measure identified was to avoid PSA screening in men aged ≥ 70 years. Most measures focused on diagnostic testing and staging (n = 3). All were related to avoiding low-value imaging (PET, CT, or radionuclide scans) for staging purposes within certain time frames after diagnosis in low-risk patients. The one treatment-specific metric encouraged counseling patients on the risks and benefits of engaging in an active surveillance program for their prostate cancer before offering procedures like interstitial prostate brachytherapy, external bean radiotherapy, radical prostatectomy, or cryotherapy. These procedures are commonly overused in the treatment of prostate cancer and may expose patients to potentially unnecessary risks.45

Discussion

This is the first article to the authors’ knowledge identifying, quantifying, and categorizing cancer quality measures promoting the avoidance of low-value care. Since healthcare quality measures were first implemented in the USA in the late 1990s, increasing the value of healthcare has been a top priority.46 Multiple studies and quality measure organizations have reported the need for quality measures targeting avoidance of low-value care to support this mission.17,47,48 In this review, we demonstrate that 18% of cancer-specific quality measures promote the avoidance of low-value care and identify gaps in phases of care where quality measures focusing on low-value care were not found.

We identified several gaps where evidence-based recommendations supporting the avoidance of low-value care are not represented in quality measures. For example, no quality measures promoted the avoidance of low-value diagnosis or staging practices for lung cancer; however, since the early 2000s, multiple institutions, including the National Comprehensive Cancer Network (NCCN) and American College of Chest Physicians, have recommended against routine brain imaging in patients with lower stages of NSCLC owing to the low rates of brain metastases in patients lacking neurologic symptoms.49,50 The Society of Thoracic Surgeons and the American Board of Internal Medicine (ABIM) have now endorsed a recommendation to raise awareness of this low-value test as part of the Choosing Wisely campaign.51 Despite these recommendations, the use of this unnecessary staging test persists, with a study demonstrating that one in eight patients in the National Lung Screening Trial with Stage IA NSCLC underwent brain imaging.49 None of these patients was found to have intracranial metastases, and all subsequently underwent the standard of care treatment.

Other areas not currently represented with low-value quality measures include prostate cancer surveillance, colon cancer staging, and breast cancer clinical outcomes. In prostate cancer, 35–70% of patients have low-risk disease and would be best managed with an active surveillance program, versus a more costly and riskier intervention (i.e., prostatectomy).52,53 The overtreatment of prostate cancer, however, remains prevalent.54 Colon cancer staging is recommended in most newly diagnosed patients with a CT of the chest, abdomen, and pelvis. PET imaging in combination with CT has been shown to significantly increase costs without improving diagnostic accuracy, but its use has been increasing.55 In breast cancer, patients whose surgical specimen after a lumpectomy procedure shows cancers cells only close to the edge of the surgical margin should not receive a reoperation.56 The number of patients receiving a reoperation after pathology shows cancer cells near the surgical margin could be a measurable clinical outcome that has been demonstrated by the American Society of Breast Surgeons to be a low-value but commonly performed service.57 These three examples also have representative Choosing Wisely recommendations,51,58,59,60 but none has a representative quality measure. Broader incorporation of these types of recommendations into quality measures could help with the dissemination of these recommendations and serve as a strategy in achieving de-implementation.

By demonstrating that a minority of quality measures promote the avoidance of low-value care, this article adds to the body of knowledge showing the US healthcare system’s current definition of quality is skewed towards the active provision of care. One of the three aims for the US healthcare system is reducing the trillions of dollars we spend yearly on low-value care61, and identifying the causal problems is a critical step to accomplishing this goal. A next step includes identifying and studying the multiple barriers that are hindering the creation and application of measures targeting low-value care. One frequently cited barrier is the difficulty to accurately measure many low-value services.17,27,48 For example, colon cancer screening at least every 10 years is accepted as being high-value care in patients aged 50–75 years,62 and compliance with this recommendation is relatively straightforward to measure using administrative or claims data. After age 75 years, the USPSTF recommends engaging in shared decision-making (SDM) with the patient prior to offering screening.62 As SDM is normally reported in the medical record, but not in claims data, SDM-focused quality measures can be far more difficult to measure.63

Another barrier to broader incorporation is the lack of consensus around what constitutes a low-value service.17 Currently, many cancer tests and treatments are deemed low value because they have associated costs or risks to the patient and provide no overall survival benefit. However, providers and patients may value different outcomes than overall survival benefit (e.g., disease-specific survival, reduction in risk of recurrence, peace of mind). For example, this has been observed in older breast cancer patients undergoing SLNB. Several sources of data demonstrate this test has no overall survival benefit.64 The Society of Surgical Oncology recommends against SLNB in women aged ≥ 70 years diagnosed with early-stage, hormone-receptor-positive breast cancer. However, this recommendation has not led to complete de-implementation of this procedure.65 Qualitative studies suggest this discrepancy could stem from older patients’ preference to accept a low-value procedure in exchange for prognostic information and peace of mind, regardless of the lack of a survival benefit.66

To help reduce the significant number of low-value services provided in the USA, the concept of quality must include not only the active provision of care, but also avoiding tests and treatments that are unlikely to benefit or could potentially harm patients. Organizations like CMS have had success reducing the utilization of some low-value services through quality measures within their value-based programs9, but these types of programs are not represented throughout most quality and accreditation bodies.67 Lastly, to improve the measurability of low-value tests and treatments, attention should be placed on the development of appropriateness measures by quality collaboratives—with outcomes centered on what is important to patients and providers.

This study has limitations that warrant consideration. First, we only included cancer types with at least ten quality measures. However, the four cancer types we included are the most diagnosed and treated in adults in the USA,68 and we feel the overall principles may be generalizable to other cancer types as they pertain to low-value care reduction. Second, this review did not apply the austere search criterion needed for a systematic review that may have identified other quality-measure-producing organizations. There may be other measures created by less commonly referenced organizations not included in this manuscript. Since the organizations we included provide most of the quality measures used by hospitals and clinics, we do not believe this is a major threat to generalizability.

Conclusion

Quality measures have been shown to be effective at changing provider practices and hospital policies, and serving as a measurement tool to track our improvement when implemented.69,70,71 Quality measure programs are also calling for the development of measures that increase the value of healthcare. Broader incorporation of quality measures promoting the avoidance of low-value care could reduce harm to patients and decrease costs.