Critical Lessons From High-Value Oncology Practices

This qualitative analysis of interview results assesses the attributes of “positive deviant” oncology practices that deliver high-quality cancer care at low total cost for testing and adoption by other practices.

Key Points

Question

What are the attributes of “positive deviant” oncology practices that deliver high-quality cancer care at low total cost?

Findings

In this analysis using mixed quantitative and qualitative methods and site visits to 7 US oncology practices with structured interview questionnaires, 13 attributes were identified that likely contributed to high-value cancer care. Five attributes most clearly distinguished oncology practice sites that ranked favorably on value.

Meaning

Attributes of high-value positive deviant oncology practices can be implemented in other care systems and their contribution to value studied.

Abstract

Importance

Cancer care is expensive. Cancer care provided by practice organizations varies in total spending incurred by patients and payers during treatment episodes and in quality of care, and this unnecessary variation contributes to the high cost.

Objective

To use the variation in total spending and quality of care to assess oncology practice attributes distinguishing “high value” that may be tested and adopted by others to produce similar results.

Design, Setting, and Participants

“Positive deviance” was used in this exploratory mixed-methods (quantitative and qualitative) analysis of interview results. To quantify value, oncology practices located near the US Pacific Northwest and Midwest with low mean insurer-allowed spending were identified. Among those, practices with high quality were selected. A team then conducted site visits to interview practice personnel from June 2, 2015, through October 3, 2015, and to probe for attributes of high-value care. A qualitative analysis of their interview results was performed, and a panel of experienced oncologists was convened to review attributes occurring uniquely or frequently in low-spending practices for their contribution to value improvement and ease of implementation. Four positive deviant (ie, low-spending) oncology practices and 3 oncology practices that ranked near the middle of the spending distribution were studied.

Main Outcomes and Measures

Thematic saturation in a qualitative analysis of high-value care attributes.

Results

From the 7 oncology practices studied, 13 attributes within the following 5 themes emerged: treatment planning and goal setting, services supporting the patient journey, technical support and physical layout, care team organization and function, and external context. Five attributes (ie, conservative use of imaging, early discussion of treatment limitations and consequences, single point of contact, maximal use of registered nurses for interventions, and a multicomponent health care system) most sharply distinguished the high-value practice sites. The expert oncologist panel judged 3 attributes (ie, early and normalized palliative care, ambulatory rapid response, and early discussion of treatment limitations and consequences) to carry the highest immediate potential for lowering spending without compromising the quality of care.

Conclusions and Relevance

Oncology practice attributes warranting further testing were identified that may lower total spending for high-quality oncology care.

Introduction

Cancer care is complex, variably delivered, and usually conducted in an ambulatory setting by physician-led teams. Care delivery teams operate at the intersection of basic biology, clinical medicine, and patient emotional needs. The complexity and variation in care delivery creates opportunities to identify attributes of ambulatory oncology practices that deliver high-value care, that is, high-quality care at a relatively low total cost. Because value-based payments—implementation of the Medicare Access and Children’s Health Insurance Program Reauthorization Act, the Oncology Care Model, and Accountable Care Organizations—are already rewarding high-value institutions, a better appreciation of the attributes of high-value oncology care is timely. Combinations of claims data with cancer registry and clinical data can broadly identify high- and low-value care, but smaller units of care delivery, such as an ambulatory oncology practice, are not well studied.

Measuring and defining value includes quantitative measures of the total cost of care and measures of quality. One measure of ambulatory cancer care is provided by the American Society of Clinical Oncology’s Quality Oncology Practice Initiative (QOPI) and the associated voluntary QOPI certification program. To identify qualitative attributes of groups that deliver high-value cancer care, we used the “positive deviance” technique. Positive deviance detects superior existing solutions that can then be adopted to produce similar results. Positive deviance has identified fresh approaches to patient activation as well as physician engagement and clinical care after myocardial infarction and has been used previously by our group.

Our exploratory hypothesis-generating study started with quantitative methods to identify positive deviant practices, meaning practices that ranked favorably compared with their peers on low mean total spending per treatment episode and that also scored highly on ambulatory quality measures. We then applied qualitative methods to identify potentially transferable attributes of high-value care.

Methods

Quantitative Methods to Select Potential High-Value Practices

The Hutchinson Institute for Cancer Outcomes Research and Anthem Inc provided data that enabled the generation of a sample of oncology practices in western Washington state and 2 Midwestern states. Practices were defined by their unique federal tax identification numbers; when the same tax identification number was associated with several sites of care, the largest site was selected. Sites, practices, practice organizations, oncologists, and oncology practices are hereinafter referred to as practices. Participation in the study was voluntary. The institutional review board at Stanford University determined that the present study was exempt from institutional review board review and waived the need for participant informed consent.

The Hutchinson Institute for Cancer Outcomes Research developed a measure of quality that included adherence to 4 of the Choosing Wisely treatment recommendations (as an indicator of quality) (eTable 1 in the Supplement), spending (including health insurance claims data from Premera Blue Cross), and tumor staging information from tumor registries. Data were used from 3225 patients whose care was attributed to 16 practices from January 2007 through May 2014. Potential high-value practices were in the lowest third of the spending distribution and were considered “wise choosers,” whereas comparator practices were in the middle third on spending. As a marker of quality, QOPI-certified practices were selected.

In another region of the United States, Anthem Inc identified 32 973 episodes of care provided by 206 practices in Ohio and Indiana from July 2010 through July 2012. Anthem Inc used the Optum Episode Treatment Group, a case-mix adjustment and episode-building system that uses routinely collected inpatient and ambulatory claims data (eg, claims for physician services, chemotherapy and associated administration, and imaging) to compare mean total spending per treatment episode among health care practices. Based on the Optum Episode Treatment Group output, potential high-value practices were identified with observed to expected costs below the mean and potential comparator organizations with cost and CIs overlapping the mean. Because Choosing Wisely performance data on these practices were not available, the American Society of Clinical Oncology’s QOPI certification program was used such that potential practices were further winnowed to those that were QOPI-certified.

In total, 11 oncology practices remained after excluding those practices that did not predominantly provide medical oncology services, those with substantial missing data (eg, omitted physician codes), or those with too few patients attributed to the practice physicians. Of these, 4 high-value practices, which were designated positive deviant practices, and 3 comparator practices were assessed in our qualitative study. Additional details are in the eMethods in the Supplement.

Site Visit Protocol

Among others, two of us (D.W.B., a senior medical oncologist, and B.P., a qualitative researcher) conducted 2-day site visits from June 2, 2015, through October 3, 2015. Physicians, advanced practice providers (including licensed nurse practitioners and licensed physician assistants), nurses, medical support, administrative office personnel, and practice leaders and managers were interviewed. A structured interview tool, designed prior to site visits and containing questions covering clinical, nonclinical, and quality management topics, was used uniformly. The interview questions were shared with practice personnel prior to the visit, and the site visitors recorded extensive interview notes. Questions probed care delivery methods, staff roles and functions, patient services provided, cultural norms, and diagnostic and surveillance testing use. The site visit team also solicited interviewee opinions regarding attributes that could affect quality and total spending. All interviewers and interviewees were blinded to practice status (ie, high value or comparator).

Qualitative Analysis of Interviews

A separate team debriefed the site-visit team within 48 hours of a site visit. The site-visit team also prepared a written report summarizing the interviews.

A 5-stage framework approach, developed for applied qualitative research, was used to identify attributes potentially responsible for high value. In the first familiarization stage, interviewee “mentions” of features contributing to high value and high-quality care were identified. (Mentions from all 7 site visits were treated equally regardless of the practice’s classification.) Based on these mentions, in the second stage, a thematic framework was identified (eg, “treatment planning and goal setting”) as given in the first column of Table 1. For stage 3, site-visit mentions were grouped into “attributes,” and each attribute was categorized into the appropriate theme to identify distinctive attributes among the practices. Each attribute and its implementation details were charted in stage 4. For the final stage, distinct attributes that might explain the high value were summarized, synthesized, and identified. Attributes were subdivided into patient- and practice-centered themes. Attributes found in low-spending practices and in comparator practices were then tallied.

Table 1. Unique, Distinguishing, and Foundational Attributes of 4 High-Value and 3 Comparator Organizations.

Theme	Attribute	Example	High Value, No. (%)	Comparator, No. (%)
Unique: Found in 3-4 (75%-100%) High-Value Organizations and 0 Comparator Organizations
Treatment planning and goal setting	Conservative use of diagnostic and surveillance imaging	Physicians use guidelines as a reference to decide whether to order diagnostic imaging and laboratory tests	4 (100)	0
		Imaging and laboratory tests only ordered if result would influence treatment plan; decision is based on physician’s clinical judgment or through a mechanism for peer review (ie, multidisciplinary tumor board)
		No imperative to “chase down every lead” and uncover reasons for every finding, in contradistinction to classic medical training
	Discussion on treatment limits and consequencesa	Care team sets expectations during initial conversations (at first or second office visit) with discussions on treatment prognosis and implications (ie, quality of life, adverse effects, and symptoms)	4 (100)	0
Support for patient journey	Single point of contact (usually a nurse) helps patients navigate the oncology care system	Each organization has ≥1 dedicated “go to” person to help patients understand their journey, care team, and available resources	3 (75)	0
Care team functions at the highest level of competence and license	Maximal use of RNs to make care decisions and appropriate interventions	RNs assess patients for adverse effects and initiate symptom management	3 (75)	0
		RNs review chemotherapy effects and recommend midcourse corrections; most nurses are OCNs
External context	Multicomponent health system–affiliated	Physicians or a physician-owned group employed by a multicomponent health system	3 (75)	0
Distinguishing: Found in Most High-Value and 1 Comparator Organization
Support for patient journey	Proactive and ongoing assessment for signs and symptoms that trigger further assessment or triage to in-office or higher level of care	Organization uses a process to routinely and proactively screen all patients for “red flags” (eg, unexpected disease or treatment complications)	2 (50)	0
		Organization conducts further or more frequent evaluation of high-risk patients
		Organization visibly tracks “saves,” including diverting patients from emergency departments to a more appropriate, often specialized, level of care, such as an urgent care facility; providing hydration in an infusion center; or securing an urgent office visit
	Palliative care incorporated early in the care arc and normalizeda	Palliative care is explained to patients and family as an integral part of treatment, is incorporated early, and is not limited to end-of-life care	3 (75)	1 (33)
		A dedicated, specialized resource supports the provision of palliative care rather than relying solely on the medical oncologist to provide palliative treatment
Care team functions at the highest level of competence and license	Patient issues solved as a team working together	Regularly scheduled multidisciplinary tumor boards	4 (100)	1 (33)
		Informal, unplanned “curbside consults” among care team members encouraged and occur frequently
	Hospitalists used to maximize medical oncologists’ office-based time	Organization provides or collaborates with a dedicated oncology-focused hospitalist	3 (75)	1 (33)
		Hospitalist’s exclusive responsibility is management of hospital inpatient care
		If a nononcologist, the hospitalist closely coordinates hospital care with the patient’s primary, office-based oncologist
Technical support and physical layout	Physical location configured for informal consultation and collaborative work	Care team works in a compact, multipurpose physical space that enhances intrateam verbal and visual communication	4 (100)	1 (33)
		Care team is adjacent to other departments (eg, imaging, laboratory, radiation oncology, and infusion center)
		Effective use of EHRs as a communication tool among all members of the care team to facilitate high-value care; routine incorporation of EHRs into regular workflow (rather than using email outside of the EHR, text messaging, etc) and using the equivalent of instant messaging within EHRs to create task lists and reminders
External context	Small size	<10 Medical oncologists in organization	4 (100)	1 (33)
Foundational: Found in Both High-Value and Comparator Organizations
Support for patient journey	Ambulatory rapid response for patients with an unstable conditiona	Organization provides rapid real-time response by phone and in office to triage patients and provides urgent care (eg, hydration) in the office or in an adjacent infusion center	4 (100)	3 (100)
	Patients with a high emotional burden or lacking in social support proactively identified, offered supportive services, and frequently and proactively reassessed	Care team has a structured approach to assess patient nonmedical needs	3 (75)	2 (67)
		Uniform initial and repeated distress assessment
		When needs are identified, patient referred to supportive resources and follow-up is initiated
Care team functions at the highest level of competence and license	APP patient care roles are maximized (“upshifted”)	APPs practice collaboratively with oncology physicians but have a high level of autonomy	4 (100)	3 (100)
		APPs have their own panels of patients in palliative care and survivorship
		As their training and license permit, APPs provide urgent care and “drop-in” visits and scheduled nonurgent follow-up visits for surveillance and symptom management
Technical support and physical layout	EHR used effectively for communication and care coordination	EHR includes embedded cancer care management tools (eg, protocols)	3 (75)	3 (100)
		EHR tools provide near real-time information and support care coordination and interprofessional communication
		Tools are used to generate task lists

Abbreviations: APP, advanced practice provider, including licensed nurse practitioners and licensed physician assistants; EHR, electronic health record; OCN, oncology-certified nurse; RN, registered nurse.

^aAttributes identified by the expert panel as having the highest potential contribution to cost and quality.

Recommendations for theory-based content analysis were followed to achieve thematic saturation, that is, identifying all themes from interview data. The practices to study and the roles for interviewers and interviewees were established a priori, and the same structured interview questions and interview guide were used for all practices. A minimum of 10 interviewees were included from each practice. The mentions and attributes were organized and presented using cumulative frequency graphs to enhance the transparency and verifiability of the decision that saturation had been achieved and to address complex or multifaceted descriptions. The analysis continued until no new themes emerged. The interview results were coded by 3 independent coders (M.K.S., B.P., and C.L.) from the debriefing team.

Review of Findings by an Oncology Expert Physician Panel

The results uncovered in the qualitative aspect of this study were further refined by a recruited expert panel of experienced oncologists. Using a modified Delphi process, the panel scored each attribute for its potential to lower mean spending per episode without compromising quality of care. A composite of their scores was then computed.

Results

Attributes of High-Value Practices

The characteristics of the 7 practices that agreed to participate and were visited by one of our site-visit teams are given in eTable 2 in the Supplement. The team identified 13 distinct practice attributes that may have affected care cost and quality. These practice attributes were grouped into the following 5 themes and are given in Table 1: (1) treatment planning and goal setting; (2) support for the patient journey; (3) care team organized so that members function at the highest level allowed by their competence and license; (4) technical support and physical layout; and (5) external context. Five “unique” attributes were found only in high-value practices, that is, none of these attributes was found in the comparator practices. Five other “distinguishing” attributes were much more frequent in high-value practices but were also found in the comparators. Three “nondistinguishing” attributes occurred at similar frequencies in both high-value and comparator practices.

Unique Attributes of High-Value Practices

Treatment Planning: A Conservative Approach to Diagnostic Testing

High-value practices used remarkably similar language to describe their approach, such as “[we order tests] to minimize [inadvertent] repetition and [based] on [the] medical necessity of ‘only if the test result will make a difference in patient care.’” Oncologists and diagnostic radiologists explicitly discussed the most efficient testing route to patient goals (eg, using the same imaging tests for both cancer staging and radiotherapy planning) and chose those diagnostic tests with the highest utility. This process was described by clinicians as a “conservative” or “less is more” approach. Adoption and adherence to this conservative testing approach was reinforced by routine group discussions and by coordinating test use during tumor boards and case conferences.

Treatment Planning: Setting Goals After Explicit Discussion on the Benefits, Limits, and Consequences

Physicians and other team members emphasized conducting discussions early after initial diagnosis, during the first or second office visit, or after significant clinical events (eg, cancer recurrence) to incorporate input from the conversations into the treatment plan and to “set realistic goals.” Sufficient time was allotted to ensure the patient understood the treatment, the available patient support services, and the availability of concurrent palliative care (curative treatment of treatable conditions, including infections; symptomatic treatment of dyspnea and pain, as well as end-of-life care). The physician and other team members revisited these discussions at follow-up visits to ensure that treatments were continuously aligned with patient goals.

Support for the Patient Journey

Proactive support for patients during predictably stressful periods (eg, cancer relapse or unexpected scan results) was also unique to the high-value practices. A single staff member was often the “point person” or “go-to person” to assist patients in understanding their disease and learning to navigate the health care system. Often trained as a nurse, this point person frequently met with the patient (usually at each office visit), offering tips on self-management and help with a range of common stressful issues, including financial and transportation assistance and access to services that provide social and emotional support.

Care Team Functions at the Highest Level of Competence and License

The use of experienced oncology nurses and other nononcologist care providers was another often-mentioned attribute. Nurses worked via protocol to provide clinical assessments and management suggestions (typically in response to inbound telephone calls); to triage nonscheduled, urgent, or emergent patient evaluation; and to offer management by nonphysician clinical staff (usually advanced practice providers, including nurse practitioners, advanced practice nurses, or physician assistants) or to nononcology physicians.

External Context

Close affiliation with a large and generally hospital-based health system or with a health plan that employed physicians was also a feature of high-value practices. However, the oncology practice unit was small and cohesive and retained its distinct identity within a larger system. The affiliated larger systems provided a broad range of staff to support functions such as quality measurement, human resources, pharmacy, navigation, and social work.

Discussion

Themes Unique to High-Value Practices

In this qualitative analysis of high-value practices, we found 5 unique attributes, which we categorized within 4 themes. Our expert oncology panel ranked 3 of these as highly likely to contribute to a practice’s high quality and low cost of care (Table 2).

Table 2. Expert Panel–Assessed Contribution of Unique Attributes to Cost and Quality^a.

Attribute	Impact Score
	Quality	Cost	Combinedb
Palliative care incorporated early in the care arc and normalizedc	4.0	4.5	8.5
Ambulatory rapid response provided for patients with an unstable conditionc	4.2	4.2	8.4
Limits and consequences of treatment discussedc	4.0	4.3	8.3
Signs and symptoms proactively and continually assessed for the need of further assessment or triage leading to in-office or higher level of care	3.7	4.0	7.7
Diagnostic and surveillance imaging used conservatively	3.0	4.2	7.2
Patient issues solved as a team working together	2.8	4.0	6.8
Nurses (RNs) maximally used to make care decisions and appropriate interventions	3.0	3.5	6.5
Hospitalist used to maximize medical oncologists’ office-based time	3.0	3.2	6.2
APP patient care roles maximized (“upshifted”)	2.7	2.8	5.5
Electronic health record used effectively for communication and care coordination	2.2	3.2	5.4
Physical location configured for informal consultation and collaborative work	2.0	3.0	5.0
Single point of contact, usually a nurse, provided to help patients navigate oncology care system	1.8	2.5	4.3
Patients with a high emotional burden or lacking in social support proactively identified, offered supportive services, and frequently and proactively reassessed	1.7	2.3	4.0

Abbreviations: APP, advanced practice provider, including licensed nurse practitioners and licensed physician assistants; RNs, registered nurses.

^aAttributes of the 7 practices ranked on a 5-point scale (5 indicates greatest impact) based on the contribution of the attribute to cost and quality as judged by an expert panel.

^bSum of the panel’s mean score of each attribute’s contribution to cost and to quality.

^cAttributes having the highest potential contribution to cost and quality.

Theme 1: Treatment Planning and Goal Setting

Physician restraint in the use of laboratory testing and imaging was often expressed as “we don’t order tests if the result won’t change the treatment plan” or shown by the preplanned use of 1 imaging procedure for both diagnostic and radiotherapy planning. In addition, global treatment planning and realistic goal setting were found to engage patients and families along the cancer journey. Engaged patients and families were considered under the care of the team. Care teams “added value by talking more.”

Theme 2: Navigation and Palliative Care in Support of the Patient Journey

Oncology care includes multiple procedures, tests, and treatments supervised by different physicians, requires transfers and exchanges among multiple locations, and necessitates communication with insurers and pharmacies. Care coordination facilitates more efficient use of time and less duplication. Care coordinator or patient navigator are terms often applied to this function. Although the value of care coordination has been difficult to demonstrate in broadly focused ambulatory care organizations, its value may be greater in oncology because of the multiple clinicians engaged in the care of each patient, and its value has been recently demonstrated in older patients with cancer.

Early introduction of palliative care services and normalizing palliative care—“this is the way we always do it”—is another high-value attribute. Normalizing palliative care mitigates the sometimes negative connotations of the end-of-life or hospice care associated with palliative care. An example of this successful attribute was the following: “[(O)ur palliative care team has] taken care of a family member of almost all of our medical staff, and they have experienced firsthand the benefits we provide.” Our methodology could not distinguish among the many aspects of palliative care to determine which had the greatest impact.

Theme 3: Care Team Functions at the Highest Level of Competence and License

Experienced, well-trained nurses performed clinical assessments and made protocol-based patient self-management recommendations. If a patient needed urgent care, these nurses would often direct patients to an on-site ambulatory care facility. This facility was generally in the chemotherapy infusion area or was to a contracted urgent care facility. The goal was to avoid unnecessary, expensive, and inconvenient emergency department visits.

In addition, advanced practice providers often provided chemotherapy symptom management, survivorship care, and palliative care. This system enabled physicians to use their skills to focus on complex clinical problems and their time to develop deeper patient relationships and to facilitate shared decision making by patients and families.

Theme 4: External Context

Although our exploratory study focused on attributes of care delivery, the smaller units of care delivery that surfaced as positive deviants on value often benefitted from administrative infrastructure (eg, a common electronic health record platform, human resources, and compliance personnel) supplied by an associated health system. The combination of “small care” and “big administrative support” may enable care teams to be nimbler in decision making, more open to adoption of best practices, or better at relational coordination.

Role of Qualitative Research in Determining Value

Efforts to measure and improve quality and increase the value in health care, such as through the QOPI, have focused on improving adherence to processes of care. Implicit in this quality improvement effort is the assumption that clinical trial outcomes (eg, improved overall survival, disease-free survival improvements, and reduced toxicity) can be translated into processes, and adherence to processes will improve care and lead to better outcomes. While process measures and guideline adherence have utility and can be quantified, they provide little guidance on potentially important and nonobvious attributes of care delivery. Qualitative research methods, such as interviews of positive deviant oncology practices, can expand our understanding of how value is created by care teams.

Limitations

The results from our small, hypothesis-generating study are insufficient to support widespread adoption of the attributes that we reported. In addition, our results are strictly applicable only to QOPI participants, who represent approximately 15% of practicing US medical oncologists. High-quality positive deviant oncology practices who do not participate in a QOPI were missed in this study. Although we achieved thematic saturation, there may be other high-value attributes that we missed. We did not explore features of palliative care that were perceived as providing the greatest impact. Our methods also did not permit exploration of other potential drivers of cost, including practice consolidation and cost increases surrounding the implementation of the Patient Protection and Affordable Care Act. We also did not explore the role of increasing costs of anticancer drugs or drug choice, although both drug costs and negotiated prices were captured in the claims data and are likely critical components of episode spending. We also did not study low-value practices nor did we independently rank costs by detailed examination of claims as reported to Premera Blue Cross or to Anthem Inc. We did not have data from a national database to compare practices on standard measures of health spending, necessitating different practice selection methods in the regions we studied. Results from the Oncology Care Model may provide an opportunity to confirm our results on a more uniformly selected group of practices. Other studies operating with different data sources may not reproduce our results, but our interview and analysis methods should be relevant. Expert panel physicians ranked attributes associated with physician involvement highly. A more diversely composed panel may have led to a different ranking. Finally, our correlation-based study does not establish causality or the relative importance of or the relationships between the attributes that correlated with low spending per episode of care.

Our results can be viewed as those from a “training” data set; organizations that choose to implement our findings should carefully study—as a “validation” set—the contribution of our results in achieving better health, quality care, and low cost. Validation should preserve the validity of our discriminatory measures (eg, standard interview questions, interviewers blinded to practice cost status, multiple practice personnel interviewed across multiple job functions, and thematic saturation) and verify and examine with more granularity the contribution to costs associated with the practice attribute. Validation might also include identifying other care practices that discriminate excellent from good to confirm the validity of the important attributes provided in Table 1. Our results will inform design of larger, confirmatory qualitative studies as well as of larger value-based data sets.

Conclusions

Organizations facing increased pressure to lower health care spending and improve quality of oncology care can view this study as an additional source of insight, until readily transferable attributes of care are tested and available to inform more refined system designs. Changes to the decision-making culture of laboratory testing and diagnostic imaging will require physician leadership and participation and administrative and implementation science support. Upgrading staff roles and improving care efficiency will require a system redesign. Patient-centered redesign of care is valuable and should be central to any change management effort. Our preliminary study has identified attributes of some of the most valuable care in the United States.

Supplement.

eTable 1. Four Quality and Cost Measures Based on 2012 Choosing Wisely Recommendations

eTable 2. Characteristics of Oncology Organizations

eMethods. Data Sources, Descriptions, and Cost Data Computation Methods