Quality improvement in neurology: Neuro-Oncology Quality Measurement Set

Over 78,000 new central nervous system (CNS) tumors are diagnosed each year in the United States, nearly one-third of which are primary malignant brain tumors,¹ and the US prevalence of primary brain tumors is approximately 688,000.² While primary CNS neoplasms represent only 1.4% of new cancer diagnoses, approximately 2.7% of cancer deaths are related to CNS neoplasms,³ and it is estimated that 16,947 deaths will result from primary CNS tumors in 2017.¹ Population-based studies have shown that socioeconomic disparities are present within the neuro-oncology community,^4–7 highlighting the need for a unified system of quality metrics in this growing field. Toward this end, several authors have published work on quality-based practice and on the inclusion of patient reported outcomes in brain tumor care.^8–12

During the launch of AAN’s Axon Registry^®, the American Academy of Neurology (AAN) requested that subspecialty societies identify gaps in subspecialty care amenable to clinical quality measure development, and work to identify ways the AAN could help meet those needs. Thereafter, the AAN and the Society for Neuro-oncology (SNO) identified a small work group to determine neuro-oncological gaps in care, to evaluate supporting evidence for clinical practice standards, and to develop feasible clinical quality measures to address these areas. The hope is that these measures will help drive clinical practice improvement and better patient outcomes.

Opportunities for Improvement

Following a thorough literature search, the work group identified five areas in need of quality improvement. The specific topics and rationale for each are described below.

Multidisciplinary Care Plan Development

Multidisciplinary tumor board discussions for care plan determination have been associated with improved quality and coordination of care in various cancers, and are a well-established quality indicator in oncology care, both domestically and internationally.^8,13–15 Indeed, establishment of a regularly meeting tumor board with input from neuro-oncology, neurosurgery, radiation oncology, neuro-radiology, and neuro-pathology has been cited as a necessary component for establishing a brain tumor center.¹⁶ One study of brain tumor board discussions revealed that 91% of 1516 clinical recommendations were implemented, and that nearly half of those were recommendations for conservative management, demonstrating the utility of multidisciplinary input in neuro-oncology cases and in optimizing patient centered outcomes.¹⁵ Another study showed that patients treated by physicians who attend weekly tumor boards are significantly more likely to be enrolled in clinical trials for various cancers, and patient survival is better when physicians attend a specialized tumor board as compared to a general oncology tumor board.¹⁷ Thus, a treatment gap does exist.

Molecular Brain Tumor Testing

Recent advances in molecular characterization of brain tumors have improved the understanding of disease pathophysiology and prognosis, and have provided new therapeutic opportunities. For the first time, the 2016 World Health Organization (WHO) Classification of Tumors of the CNS includes molecular parameters in addition to histology, identifying integrated diagnoses as the new standard for maximal diagnostic specificity.¹⁸ Among the molecular testing recommended for gliomas, isocitrate dehydrogenase (IDH) mutation analysis and 1p/19q co-deletion reporting are required for integrated diagnoses. It is hoped that the use of integrated diagnoses will facilitate further progress in research and therapeutic efficacy for brain tumor patients. The work group believes a treatment gap exists as providers and hospitals work to implement the most recent WHO guidelines.

Chemotherapy Education and Informed Consent

American Society of Clinical Oncology and Oncology Nursing Society guidelines indicate that all patients who are prescribed chemotherapy should be provided education in advance of prescription.¹⁹ The most frequently used chemotherapies in neuro-oncology are oral, including temozolomide, lomustine, and procarbazine, and errors in the home administration of oral chemotherapy are associated with a high likelihood of harm.²⁰ Education of patients and caregivers to reduce the risk of error is vital, as well as to empower them to speak up if an error occurs.²⁰ Further, a patient’s (or proxy’s) consent to treatment after receiving adequate education is an important part of delivering quality cancer care, and is not only ethical but may mitigate legal risk in some cases.²¹ Written documentation of consent is stressed among oncology quality best practices,²¹ which may be either in the form of documentation by the clinician or a document signed by the patient or proxy. A review of practice-level systems indicated opportunities for quality improvement efforts for the safe management of chemotherapy, defining the treatment gap targeted here.²²

Peri-Operative Magnetic Resonance Imaging (MRI)

For high-grade gliomas, extent of resection is strongly correlated with overall survival.^23,24 Furthermore, many clinical trials include measures of residual disease as a criterion for inclusion. Post-operative imaging is a requisite for radiation oncologists to design both safe and effective treatment plans, and finally, radiographic monitoring of therapeutic efficacy is contingent upon a baseline (post-operative, pre-therapy) MRI scan. Given vascular and inflammatory changes that occur after brain tumor resection, which can reduce imaging specificity, guidelines exist to perform a postoperative MRI scan with and without contrast within 24–72 hours after resection.^25,26 In fact, findings from early post-operative MRI itself have been directly shown to correlate with survival.²⁷ Intraoperative MRI may also be used during tumor resection, though gadolinium is often not given intraoperatively and this lack of contrast material may reduce the utility of the imaging for the above referenced items. A retrospective study of a single neurosurgeon shows that intraoperative MRI-based estimation of gross total resection vs subtotal resection was accurate 79.6% of the time compared to gold standard post-operative MRI, and estimates of extent of resection improved from 72.6% to 84.4% over a 17 year period, highlighting the learning curve of intraoperative MRI resection estimations.²⁸ To the best of the authors’ knowledge, no other head-to-head comparison exists between intraoperative and post-operative MRI. Current statistics on performance rates of intraoperative or post-operative gadolinium enhanced MRI for glioma are not available. It is anticipated that by measuring performance, a treatment gap will be confirmed and further opportunities for improvement will be identified.

Venous Thromboembolism

Venous thromboembolism (VTE) is a common complication of systemic malignancies, and is the second leading cause of death among patients with cancer.^29,30 Patients with brain tumors, especially glioblastoma, are at an increased risk of VTE compared to other types of cancer.^31,32 While trials of long-term prophylactic anticoagulation have not demonstrated significant reductions in VTE risk among high grade glioma patients,³³ perioperative inpatients are considered to be at especially high risk of VTE due to the addition of both immobility and systemic inflammation to an already pro-thrombotic oncologic state. Compression stockings are generally used as a low risk option for VTE prevention peri-operatively, yet high rates of VTE are seen even with their use.⁹ Notably, two prospective clinical trials have shown safety and improved efficacy in adding low molecular weight heparins to pneumatic devices,^34,35 though the majority of patients still do not receive chemoprophylaxis after brain tumor surgery.³⁶ Based on the strong data regarding VTE prophylaxis, yet the immeasurable complexity of provider decision for any one patient, an outcome measure was proposed to determine the actual rate of inpatient perioperative VTE development. Since approval of the measure specifications, Senders et al. assessed VTE rates following craniotomy for a primary malignant brain tumor through extraction of National Surgical Quality Improvement Program (NSQIP) data.³⁷ The work group hopes that through implementation of the developed quality measure real-time data will be available for clinicians to assess and address VTE rates following resection and biopsy for grade III-IV gliomas.

The work group explored multiple alternate measure topics (Appendix e-1) addressing various other practice patterns and clinical processes. Ultimately these measures were not developed for a variety of reasons, including difficulty capturing data due to current electronic medical records practices, lack of evidence, lack of a known gap in care practice, and development of similar measures by others in related fields. These concepts will be retained for future measurement set updates, as more evidence may support development at that time.

Methods

The AAN is an established measure developer and steward of quality measures for neurologists. Details of the AAN’s full measure development process are available online.³⁸ AAN and SNO worked in partnership to develop a de novo set of quality measures for neuro-oncology, and in so doing, piloted an expedited, ‘virtual’ version of the measure development process. This expedited process included simultaneous completion of an evidence-based literature search by a medical librarian and identification of members of the multi-disciplinary work group, all of whom adhered to the AAN conflict of interest policy. Work group members were selected by the AAN’s Quality and Safety Subcommittee (QSS) measure development leadership team through a process that reviewed relevant clinical experience, prior exposure to or work in quality measure development and/or quality improvement, and potential conflicts. The final work group comprised nine members representing multi-stakeholder interests for patients with neuro-oncology conditions. Unfortunately, the call for work group members did not result in identification of any potential patients and/or caregivers. One QSS member served as a non-voting methodological facilitator.

Once the work group was constituted, members drafted candidate measures with attendant technical specifications, identified measures ripe for development via a modified Delphi process, and refined candidate measures via teleconference meetings. The work group focused their efforts on measures that arose from known treatment gaps (allowing for opportunity to improve) with evidence or guideline statements to support existing standards of care, had high face validity, and were feasible to collect. Candidate measure concepts were extensively reviewed and edited prior to a work group vote to approve, reject, or abstain on each measure during teleconference meetings.

Subsequent to this preliminary work group approval, a 21-day public comment period resulted in input from 22 individuals, a similar number to public input for previous measure development efforts. Efforts were made to publicize this opportunity both to patients and to caregivers in the interest of identifying meaningful measures. Following public comment, measure concepts were further refined prior to measurement set approvals by the work group, the AAN QSS, the SNO Guideline Committee, the AAN Practice Committee, the SNO Board of Directors, and the AAN Institute Board of Directors.

The AAN and SNO will update these measures as needed every three years, allowing the measurement set to provide a working framework for measurement, rather than a long-term mandate.

Results

The workflow of group measure proposal, evaluation, and development is shown in Fig. 1. The process began with a literature review by a medical librarian, identifying 1,052 relevant abstracts, 25 potential guidelines or systematic reviews, and 60 articles providing supporting evidence. After work group members reviewed the literature outlining care gaps and AAN quality measure requirements (available online at https://www.aan.com/practice/quality-measures/ and in Appendix e-2), the group proposed 15 initial concepts for evaluation (Appendix e-1). Three items that were duplicative of other national organizations’ quality measures were omitted. Then, a second literature review focused on level of evidence for the remaining proposed measures, and four additional measures were omitted for insufficient evidence. Finally, group discussion determined that electronic medical record data extraction for three more measures would be prohibitively challenging for clinicians and organizations, and so these were omitted. In the end, the group delineated five areas that were feasible for measure development, as they had support from scientific literature and evidence of gaps in care. These concepts were refined through an iterative development process, submitted for public comment, and ultimately approved by the work group and governing bodies of AAN and SNO (Table 1). While four of these measures focus on the process of providing evidence-based neuro-oncological care, the fifth measure is an outcome measure, determining the incidence of VTE after high-grade glioma resection or biopsy. The work group felt that, while the use of post-operative VTE prophylaxis was nearly ubiquitous, practice patterns regarding mechanical or chemical VTE prophylaxis may vary by provider or be influenced by patient factors that cannot be easily captured through electronic medical record. As such, the work group determined that measuring the outcome of post-operative VTE incidence would be the most effective gauge of quality neuro-oncological care. It should be noted that screening for VTE in asymptomatic patients is not recommended; this outcome measure is based on VTE identification on imaging performed for clinical suspicion.

Fig. 1.

Measure Development Flow Diagram. Illustration of measure proposals, discussion, research, evaluation, and approval

Table 1.

2017 AAN and SNO Neuro-Oncology Quality Measurement Set. The following measures were approved by the work group. Providers are encouraged to identify the one or two measures that would be most meaningful for your patient populations and implement these measures to drive performance improvement in practice

2017 Neuro-Oncology Measurement Set
Title	Numerator	Denominator	Exclusions
Multidisciplinary Care Plan Developed for Primary Brain or Spine Tumors	Patients with a multidisciplinary tumor board treatment plan* developed in the 12-month measurement period. * See appendix e-2 for definition	All patients diagnosed in the measurement period with a new diagnosis of primary brain or spine CNS WHO Grade II-IV tumor.	None
Molecular Testing in Accordance with World Health Organization Classification of Tumors of the Central Nervous System	Patients with molecular testing performed in accordance with most recent WHO classification flow.	Patients 18 years and older who have had an initial resection or biopsy of grade II-IV glioma.	Patients with tissue insufficient for molecular testing.
Chemotherapy Education and Informed Consent for Brain Tumor Patients	Patients who were provided chemotherapy education AND for whom informed consent was obtained prior to prescription of chemotherapy.	Patients 18 years and older with a diagnosis of brain tumor who were prescribed chemotherapy outside of a clinical trial.	Education and active consent obtained for same chemotherapy prescription in the past 12-month period.
Intra-Operative or Post-Operative MRI for Gliomas	Patients who had an intra-operative or post-operative^ gadolinium enhanced MRI. ^Post-operative MRI must be obtained within 72 hours of the surgical resection.	Patients 18 years and older with a diagnosis of grade III-IV glioma who undergo a surgical resection.	-Patients with a contraindication for MRI. (i.e., presence of pacemaker, intracranial metal clip, metallic body in the eye, or neurostimulator). -Patients with a contraindication for receiving gadolinium. -Patients who undergo surgery for purposes other than cytoreduction (i.e. diagnostic biopsy only).
Venous Thromboembolism Events Following Primary Brain Tumor Surgery (Outcome)	Patients who experience a DVT or PE event during their immediate post-surgical hospitalization.	All patients who had resection or biopsy of grade III-IV glioma.	DVT or PE present admission

The work group also identified five previously developed measures complementary to this effort (Table 2) that are endorsed here for consideration in the delivery of neuro-oncological care.

Table 2.

Relevant Cancer and Advance Care Planning Measures In addition to the above AAN and SNO measures the work group recommends providers and practices consider using the following measures, which are already available for use in the field

NQF #0210 Proportion of patients who died from cancer receiving chemotherapy in the last 14 days of life.* Available at: http://www.qualityforum.org/QPS/0210

NQF #0211 Proportion with more than one emergency room visit in the last 30 days of life.* Available at: http://www.qualityforum.org/QPS/0211

NQF #0213 Proportion of patients who died from cancer admitted to the ICU in the last 30 days of life.* Available at: http://www.qualityforum.org/QPS/0213

NQF #0215 Proportion of patients who died from cancer not admitted to hospice.* Available at: http://www.qualityforum.org/QPS/0215

NQF #0216 Proportion of patients who died from cancer admitted to hospice for less than 3 days.* Available at: http://www.qualityforum.org/QPS/0216

NQF #1641. Hospice and Palliative Care – Treatment Preferences.^ Available at: http://www.qualityforum.org/QPS/1641

AAN Inpatient and Emergency Neurology Discussion and Documentation of Advance Directives.^ Available at: https://www.aan.com/practice/quality-measures/

*The work group determined there is value in using measures at either an individual or practice level. Currently endorsed by the NQF for use at a provider level, the above measures are also available and specified for use at the individual provider level.

^The work group reviewed the numerous, existing palliative and end-of-life measures already available for use in the field. The work group declined to create a new measure, and encourages providers to identify a measure to meet their population needs. Every neuro-oncology patient should be engaged in a discussion on advance care planning and research indicated patients are not engaged in these conversations early in the course of treatment. The AAN is also developing a similar Advance Care Planning measure for use in outpatient settings. The measure once finalized will be available at: https://www.aan.com/practice/quality-measures/

From the time of the initial meeting to Board approvals, development of this measurement set took ten months (July 2016 to April 2017). This compares to the historical average time of 18 months for completion of the AAN measure development process. Based on measurement set approval by two professional societies with a marked decrease in turnaround time, similar number of measures developed, and similar amount of public commentary, the work group determined the expedited, virtual measure development process to be an appropriate surrogate for previously published processes.

Conclusion

The goal of quality measures is to guide their users to evidence-based improvements in care and, eventually, health care outcomes. It is the hope of this work group that implementation of the new neuro-oncology measures will lead to objective improvements in the care of patients diagnosed with CNS tumors. There is no requirement that all measures be used by a provider or hospital. In fact, implementation of all measures at one time would likely not be feasible for providers. Instead, providers, practices, and hospitals are encouraged to identify the one or two measures that would be most meaningful for their patient populations. After collecting data, providers are encouraged to use data to drive performance improvement in practice.

Public comments and work group discussions addressed potential unintended consequences, including rising care costs and use of clinically unwarranted procedures. Commenters noted risks of manipulating data or changing practice patterns in order to avoid the perception of failure for various measures. The work group created these measures with the primary intention of improving clinical practice, understanding that providers act with best intentions to do no harm. The inclusion of an outcome measure in this set creates an opportunity for improvement even for high-volume neuro-oncology institutions that might otherwise achieve the named process measures nearly perfectly. However, perfection is not anticipated for any measure; instead, performance data should serve as an internal benchmark for provider, practice, and hospital quality improvement opportunities.

This data may highlight practice changes that providers can make to improve quality of patient care (e.g., using nurse educators to provide chemotherapy education or better documentation of informed consent procedures). Should performance rates near perfect compliance, the work group will evaluate the continued need for these measures during the next scheduled update. Future measures from this group may also include the provision of patient reported outcomes in routine clinical practice, which may inform better decisions by patients and providers.¹² Although there are no immediate plans to implement these measures in Center for Medicare and Medicaid Services’ (CMS) Merit-based Incentive Payment System (MIPS) or private payers’ performance tracking systems, it may occur in the future. Measures may be considered for use in AAN’s Axon Registry, and may also be developed for use as Improvement in Medical Practice modules for American Board of Psychiatry and Neurology Maintenance of Certification requirements. Measures will be periodically evaluated every three years and updated as necessary to reflect continued utility in quality neuro-oncology care.

Supplementary Material

Supplementary data are available at Neuro-Oncology online.

Author Contributions

1. Dr. Jordan contributed to study concept and design, acquisition of data, analysis and/or interpretation of data, drafting/revising the manuscript, critical revisions of the manuscript for important intellectual content, and study supervision including responsibility for conduct of research and final approval.

2. Dr. Sanders contributed to study concept and design, acquisition of data, analysis and/or interpretation of data, drafting/revising the manuscript, critical revisions of the manuscript for important intellectual content, and study supervision including responsibility for conduct of research and final approval.

3. Dr. Armstrong contributed to study concept and design, acquisition of data, analysis and/or interpretation of data, and drafting/revising the manuscript.

4. Dr. Asher contributed to study concept and design, acquisition of data, analysis and/or interpretation of data, and drafting/revising the manuscript.

5. Ms. Bennett contributed to study concept and design, acquisition of data, analysis and/or interpretation of data, drafting/revising the manuscript, and study supervision including responsibility for conduct of research and final approval.

6. Dr. Dunbar contributed to study concept and design, acquisition of data, analysis and/or interpretation of data, and drafting/revising the manuscript.

7. Dr. Mohile contributed to study concept and design, acquisition of data, analysis and/or interpretation of data, and drafting/revising the manuscript.

8. Dr. Nghiemphu contributed to study concept and design, acquisition of data, analysis and/or interpretation of data, and drafting/revising the manuscript.

9. Dr. Smith contributed to study concept and design, acquisition of data, analysis and/or interpretation of data, and drafting/revising the manuscript.

10. Dr. Ney contributed to study concept and design, acquisition of data, analysis and/or interpretation of data, drafting/revising the manuscript, critical revisions of the manuscript for important intellectual content, and study supervision including responsibility for conduct of research and final approval.

Disclosures

1) Dr. Jordan has no disclosures.

2) Dr. Sanders serves as a member of the National Quality Forum Palliative and End-of-Life Standing Committee.

3) Dr. Armstrong reports no current conflicts; prior to October 30, 2016, served as consultant for ABBvie Pharmaceuticals, Tocagen, Pfizer, and Immunocellular Therapeutics, and received funding from Genentech and Merck.

4) Dr. Asher has no disclosures.

5) Ms. Bennett has no disclosures.

6) Dr. Dunbar has no disclosures.

7) Dr. Mohile has received consulting fees from Novocure.

8) Dr. Nghiemphu has no disclosures.

9) Dr. Smith has no disclosures.

10) Dr. Ney has no disclosures.

Disclaimer: This article is published simultaneously in Neuro-Oncology and Neurology. Approved by the AANI and SNO work group on February 9, 2017. Approved by AANI Quality and Safety Subcommitee on February 17, 2017, AANI Practice Committee on March 8, 2017, and AANI Board on March 21, 2017. Approved by SNO Guidelines Committee on March 22, 2017 and SNO Board of Directors on April 10, 2017.