Identifying organization-level features important to patient outcomes in cancer care: qualitative interview results with experts Authors: Jennifer Elston Lafata, Soohyun Hwang, Erica S. Breslau, Melanie C. Chansky, Grace C. Huang, Annie Sampson, Sarah Kobrin Published: 2026 ABSTRACT BackgroundAddressing quality and equity requires understanding organization-level features that affect patient outcomes. Using cancer care as the context, we aimed to identify organization-level features important to patient-level outcomes and assess the feasibility of including them in clinical data warehouses.MethodsIn-depth interviews were conducted with 12 experts, including academicians with expertise in organizational theory and implementation science, oncology practitioners, and clinical data warehouse developers. Members of the research team developed a semi-structured interview guide to identify organization-level features perceived as important to patient outcomes, and the measurability and obtainability of identified features. Interviews were audio-recorded and transcribed verbatim. De-identified transcript data were organized using NVivo 12 and analyzed using thematic analyses.ResultsAnalysis revealed key organization-level features important to care quality, including leadership structure and culture, quality monitoring processes, external incentives, and support service provision. Respondents highlighted the importance of intentionality to equity, and how many of the same organization features affecting quality also affect equity. Some features were deemed readily knowable, while others presented measurement challenges to extract from existing data resources or to collect de novo. Given their ownership of the data in the US, collaboration with healthcare organizations will be essential for advancing organization-level measurement in the US.ConclusionsUnderstanding organization-level features is vital for enhancing care quality and promoting equity in cancer care and otherwise. Leveraging publicly available information and fostering partnerships between healthcare organizations and researchers could facilitate the compilation of relevant data. External incentives and measurement development initiatives could afford further advancement. Findings contribute to the understanding of organization-level determinants of cancer care quality and equity, offering practical avenues for research, data collection, and improved care and outcomes.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12913-026-14877-5. Contributions to the literature • The study addresses a critical gap in the literature by identifying key organization-level features thought to be essential for enhancing patient-level quality and equity outcomes in the context of cancer care. • While some features are readily measurable, others present challenges in data extraction or collection, emphasizing the need for collaboration with healthcare organizations to advance organization-level measurement. • The paper contributes practical steps for moving the field forward, including leveraging publicly available information, fostering partnerships, and incentivizing measurement development, thereby offering avenues for research, data collection, and improved care and outcomes in oncology. Introduction Comprehensive improvements in the health and well-being of diverse populations are unlikely without considering the multiple and interdependent levels of influence in which people live and receive care [1]. Historically, health services researchers and others have focused on the patient-treatment dyad to study patient-level health outcomes. Yet, patient outcomes occur nested within systems of influence. The levels that comprise these systems may directly or indirectly influence each other and resulting patient outcomes. As such, to understand the degree to which care receipt leads to desired health outcomes requires the ability to characterize multiple levels of contextual factors, the interplay among them, and their impact on individuals and populations health and equity outcomes. Clinical Data Research Networks (CDRNs) and emerging proprietary databases [2, 3] allow access to inter-institutional, patient-level electronic health record (EHR) data. In the context of cancer care, current efforts are underway to enable the addition of patient reported outcomes (PROs) and to use artificial intelligence (AI) to enhance data capture within these clinical data warehouses [4–6]. In addition, the linkage of community-level contextual information has grown [7]. While each of these efforts enhances the data infrastructure available for understanding the relationship between service receipt and health outcomes, to our knowledge no effort is yet underway to link organization-level features to such clinical data warehouses, thus limiting the ability to use them to address many health challenges, including those that drive disparities and support equity. To address this gap, using the context of cancer care, we used in-depth interviews to identify experts’ perceptions of organization-level features—the features that characterize an organization, whether internal (like management practices, staffing, or culture) or external (like community partnerships or regulatory policies) residing–that are important to patient-level outcomes as well as feasible to compile across diverse organizations. We did so as an initial step to identify implementation strategies for moving the field forward. Cancer care provides a rich and diverse clinical context in which to explore organization-level features and their impact on patient outcomes. Its complexity and multidisciplinary nature require attention to care coordination—often across as well as within a given clinical setting. In addition, the importance of timely care as well as the need for ongoing surveillance combines the need for rapid access to urgent services with continual access to planned follow-ups. As such the cancer care context affords aspects of both acute and chronic care management. Furthermore, a cancer diagnosis is often accompanied by diverse non-medical and emotional needs. These might include the need for financial, transportation, housing, and other supportive services such as psychosocial assistance from social workers, counselors, or palliative care and hospice care teams. Methods Interview guide development Members of the research team with expertise in cancer care delivery, health services research, and public health, developed the semi-structured interview guide. The guide consisted of a series of short, open-ended questions with probes addressing what organizational features (1) affect cancer care quality and patient outcomes, (2) affect the equity of programs designed to improve cancer care quality and patient outcomes, (3) they would like to have access to given their current professional role as well as (4) what the barriers are to obtaining these data, and (5) which of these features they thought would be obtainable from most health care organizations. The guide is available in the appendix and was designed for the interview to take approximately 30 min to complete. As indicated in the guide, participants were asked to focus on organization-level characteristics (e.g., leadership, organizational structure, policies/incentives, delivery system design, clinical decision support, clinical information systems, patient education and navigation) that were important to the effectiveness of interventions or programs where the organization level was nested between the provider and healthcare team, and the local community environment [1]. Study participants We recruited a purposely identified sample of twelve experts who collectively represented three professional groups: (1) academic experts on organizational theories, delivery system science and implementation science frameworks; (2) oncology practitioners who direct delivery of cancer care; and (3) professionals overseeing development of clinical data warehouses. The decision to include 12 experts was guided by our desire to capture diverse perspectives within the constraints of available resources. To identify participants, we used the professional networks of research team members. We determined a person’s expertise by rank (Associate or Full Professor) and peer-reviewed publications for those in academic positions, and title for those in non-academic positions (e.g., Director, Vice President, or Senior Consultant). Recruitment was stratified by professional groups to ensure the recruitment of four experts within each of the three areas. The study protocol was reviewed and approved by the Westat Institutional Review Board (FWA 00005551). Each participant provided consent to be interviewed and for their session to be recorded. Interview participants received a $50 e-gift card in appreciation of their time. Participant recruitment and data collection In the fall of 2021, two team members (JEL and SK) jointly sent introductory emails to the identified experts. A total of 23 people were emailed to generate the panel of experts. Those who explicitly declined noted prior commitments prevented them from allocating time towards the interviews. The initial email introduced the study and explained that participation would involve one 30-minute interview. Those interested in participating were asked to acknowledge their interest via reply email. Those who replied were contacted by a member of the Westat study team to obtain consent, and to arrange an interview date and time. Interviews were conducted by a trained member of the research team (MC) with a scribe present to take notes. After describing the study objectives and obtaining consent, the interviewer solicited background information from each participant (e.g., job title, type of organization, primary roles/responsibilities, and years of experience). The interviewer next shared a figure illustrating seven levels of influences on cancer care, including the individual patient, family and social support, provider/team, organization and/or practice setting, local community environment, state health policy environment, and national health policy environment, using the figure to focus the respondent on the organizational level [1]. All interviews were conducted between 10/21 and 1/22, using a video conference platform. Verbatim transcripts were professionally generated from each interview, except for one interview (participant ID 1003) for which a malfunction resulted in the absence of audio data. For this interview we used only scribe notes during analyses. All identifying information was removed from transcripts prior to analyses. After the interview, we did not recontact study participants. Codebook development and data organization We used a combination of deductive and inductive approaches to develop an initial set of domains and constructs. We began with codebook templates from the Consolidated Framework of Implementation Research (CFIR), [8] focusing on the outer setting, inner setting, and characteristics of individuals, domains and constructs. Four members of the research team (ESB, JEL, MCC, and SK) independently and then collectively reviewed four transcripts, employing an iterative, grounded, and consensus building approach for identifying emergent domains and constructs for the codebook. This process added several constructs to the coding scheme, including constructs within the inner setting, such as (1) use of choice architecture to influence decisions, and (2) quality monitoring and planning. We also added domains not specifically considered by CFIR, including (1) measurement and measurement challenges and (2) care equity. The codebook was piloted with two randomly selected transcripts. The final codebook was used to code each of the 12 transcripts. Because of topic complexity, we elected to independently code each transcript twice (i.e., double coding). Coding was done by three members of the research team (GCH, MCC, and a third trained research staff member). Discrepancies were discussed, and as needed, resolved by the third team member who did not conduct the original coding. Coded data were organized using NVivo 12 Plus [9] enabling us to generate topical reports for further in-depth analysis and the next stage of data immersion and thematic analysis. Thematic analyses The team engaged in iterative cycles of data immersion to discuss insights, resolve discrepancies, and work toward a shared understanding [10]. For this process, we organized topical reports into three overarching categories: Quality Drivers, Equity Drivers, and Measurement. Throughout this process, emergent themes were discussed among team members (AS, ESB, JEL, MCC, SH, and SK). Our approach to this thematic process evolved with the publication of a new framework from Weaver et al., 2022 [11] that added perspective to constructs already considered by the research team [12, 13]. Through the data immersion process, each emergent theme was paired with supporting quotes that provided additional clarity and poignance to the findings. Comparisons across respondent expertise groups were made but revealed no distinct patterns. Results Participant sample Twelve content experts, 7 men and 5 women, participated in an interview (52.2% participation rate): four experts each in (1) the development of clinical data warehouses, (2) cancer care delivery, and (3) implementation science or organizational theory (Table 1). Among these, four were practicing oncology care providers. Table 1Participant characteristicsID NumberExpertiseDescription of Disciplinary Expertise1002Implementation Science, and Organization TheoryPhD, Executive Director1006DBA, Professor1009PhD, Associate Professor1010PhD, Associate Professor1004Cancer CareDeliveryMD, Division Chief and Professor1008MD, Chair and Professor1007MD, Director1011MD, Associate Professor1001Clinical Data Warehouse DevelopersPhD, Associate Professor1003PhD, Chief Analytics Officer and Professor1005PhD, Chair and Co-Director1012PhD, Chief Data Officer Table 1Participant characteristicsID NumberExpertiseDescription of Disciplinary Expertise1002Implementation Science, and Organization TheoryPhD, Executive Director1006DBA, Professor1009PhD, Associate Professor1010PhD, Associate Professor1004Cancer CareDeliveryMD, Division Chief and Professor1008MD, Chair and Professor1007MD, Director1011MD, Associate Professor1001Clinical Data Warehouse DevelopersPhD, Associate Professor1003PhD, Chief Analytics Officer and Professor1005PhD, Chair and Co-Director1012PhD, Chief Data Officer Participant characteristics Cancer Care Delivery Organization-level features important to cancer care quality Leadership structure and culture Participants detailed how leadership structure and culture affect cancer care quality. For example, one participant noted how “It tends to be leaders who value a commitment to evidence-based cancer care and [who] create that culture. They reward that. They communicate that’s a value. They communicate that providing quality of care is something that is a priority and then they’re accountable. They motivate people to do it. They provide support and training. They provide this evaluation monitoring feedback…” [1008] Other participants expanded on this idea stating how it was “critically important to understand who’s controlling the cancer center product line” [1005] and highlighting the importance of that person’s authority over quality. As such respondents highlighted the importance of the reporting structure to care quality while recognizing where accountability for quality resides within the organization as important along with the values held by those organizational quality leaders. Participants further noted the importance of not only senior leaders, but also middle managers. Table 2Measurement-related themes and illustrative quotationsThemeIllustrative QuotesSome Organization-Level Features Are Currently Knowable…having locations to serve patient populations in different areas, patients who may or may not be able to travel to your main campus site… Many organizations have multiple satellite sites where they deliver care, whether the satellite sites are all located in wealthy neighborhoods or whether there’s a mix to deliver care in multiple different patient populations…that’s important to measure. [1007]… are you associated with the medical school would be black and white, or do you have a nursing school? That would be very knowable. … Who owns you? Are you owned by something? And then by some other entity, and when did that start? Very knowable. [1011]From a leadership perspective, you could always look at boards of directors of the organizations. You could always look at sort of C-suite leadership… The quality team might be a bit harder to get because it’s going to be lower down the org[anization] chart. And so, you might not have access to that info[rmation], but at least at the highest level, some of the high-level leadership, you would be able to get an understanding of diversity experience expertise within that. And I think that data could be collected pretty easily. [1012]Well, demographic characteristics generally are not that difficult to get. Payment characteristics are not too difficult to get, although it’s not easy. So, yeah. I wouldn’t say that they’re necessarily easy to collect, but there are good data sources. [1010]…there are some things potentially that could be measured indirectly, such as, using the navigator example, number of patients or percentage of your patients that are navigated or meet with a navigator on an annual basis. And that doesn’t necessarily directly give information about a navigation program, but it might indirectly. If it’s a low percentage, for example, it may not be a robust enough navigation program. [1004]Some Important Organization-Level Features Would Present Challenges…organizational culture is also really important, and it’s also difficult to collect … [service line] structure makes a difference. It is very hard to collect data on it because there are no commonly accepted terms. … One of my colleagues calls it a semantic swamp. [1006]…leadership factors, of course, are going to be probably the most difficult…very difficult to measure and to quantify leadership. [1010]…the soft touchy stuff, like how burnt out are people? How do people feel about the organization, satisfaction? Those kinds of things, I think, are hard because also we would lie to you in a survey. So I don’t even think it is worth doing…You won’t get a great representation. You’ll get people that are mad, people that love it and a lot of people who just aren’t going to fill [a survey] out or maybe they don’t want to complain about their organization. [1009]For cancer, the continuum is so important, …[the] operating linkages between different stages or organizations doing different work along the cancer continuum. … then operationally, how smoothly do they work together? Is information effectively being passed? Because, you know, at every point in those transitions there are substantial opportunities for fragmentation of care. [1006]We don’t know who we don’t take care of, necessarily, and so it’s a bit harder to measure whether we are providing equitable care. [1004]You could design surveys in particular ways that you could structure some of the answers that you want to get, that you would not be able to get from currently available data sources. … is this more of an exercise where we need to know what it is that we need to know and if it involves reaching out to individual systems and serving them to get an understanding of, for example, do they have some type of an equitable or care policy, or do they use race or social determinants as decisions in terms of to guide treatment or care? [1012]Need to Consider Willingness and PrioritizationThe other thing which I will say … will be hard is, they [the organization] may not want to give you all of this data … people are going to be really nervous about giving their financial stuff… people are going to be a little nervous about giving things that relate to power and structure. [1009]You could [not] get any data on performance evaluations, employee performance evaluations. You probably couldn’t get that kind of data. [1010]Well, the easiest to collect are things that are knowable, and organizations don’t feel could compromise their brand by sharing them. That the organizations would be comfortable associating with their brand. [1011]There are regulatory burdens that these healthcare facilities have to report. So what are they regularly reporting to Health and Human Services that you would be able to have access to? What other types of, for facility-level surveys, that they might be commercial collected or publicly available datasets are there? [1005] Table 2Measurement-related themes and illustrative quotationsThemeIllustrative QuotesSome Organization-Level Features Are Currently Knowable…having locations to serve patient populations in different areas, patients who may or may not be able to travel to your main campus site… Many organizations have multiple satellite sites where they deliver care, whether the satellite sites are all located in wealthy neighborhoods or whether there’s a mix to deliver care in multiple different patient populations…that’s important to measure. [1007]… are you associated with the medical school would be black and white, or do you have a nursing school? That would be very knowable. … Who owns you? Are you owned by something? And then by some other entity, and when did that start? Very knowable. [1011]From a leadership perspective, you could always look at boards of directors of the organizations. You could always look at sort of C-suite leadership… The quality team might be a bit harder to get because it’s going to be lower down the org[anization] chart. And so, you might not have access to that info[rmation], but at least at the highest level, some of the high-level leadership, you would be able to get an understanding of diversity experience expertise within that. And I think that data could be collected pretty easily. [1012]Well, demographic characteristics generally are not that difficult to get. Payment characteristics are not too difficult to get, although it’s not easy. So, yeah. I wouldn’t say that they’re necessarily easy to collect, but there are good data sources. [1010]…there are some things potentially that could be measured indirectly, such as, using the navigator example, number of patients or percentage of your patients that are navigated or meet with a navigator on an annual basis. And that doesn’t necessarily directly give information about a navigation program, but it might indirectly. If it’s a low percentage, for example, it may not be a robust enough navigation program. [1004]Some Important Organization-Level Features Would Present Challenges…organizational culture is also really important, and it’s also difficult to collect … [service line] structure makes a difference. It is very hard to collect data on it because there are no commonly accepted terms. … One of my colleagues calls it a semantic swamp. [1006]…leadership factors, of course, are going to be probably the most difficult…very difficult to measure and to quantify leadership. [1010]…the soft touchy stuff, like how burnt out are people? How do people feel about the organization, satisfaction? Those kinds of things, I think, are hard because also we would lie to you in a survey. So I don’t even think it is worth doing…You won’t get a great representation. You’ll get people that are mad, people that love it and a lot of people who just aren’t going to fill [a survey] out or maybe they don’t want to complain about their organization. [1009]For cancer, the continuum is so important, …[the] operating linkages between different stages or organizations doing different work along the cancer continuum. … then operationally, how smoothly do they work together? Is information effectively being passed? Because, you know, at every point in those transitions there are substantial opportunities for fragmentation of care. [1006]We don’t know who we don’t take care of, necessarily, and so it’s a bit harder to measure whether we are providing equitable care. [1004]You could design surveys in particular ways that you could structure some of the answers that you want to get, that you would not be able to get from currently available data sources. … is this more of an exercise where we need to know what it is that we need to know and if it involves reaching out to individual systems and serving them to get an understanding of, for example, do they have some type of an equitable or care policy, or do they use race or social determinants as decisions in terms of to guide treatment or care? [1012]Need to Consider Willingness and PrioritizationThe other thing which I will say … will be hard is, they [the organization] may not want to give you all of this data … people are going to be really nervous about giving their financial stuff… people are going to be a little nervous about giving things that relate to power and structure. [1009]You could [not] get any data on performance evaluations, employee performance evaluations. You probably couldn’t get that kind of data. [1010]Well, the easiest to collect are things that are knowable, and organizations don’t feel could compromise their brand by sharing them. That the organizations would be comfortable associating with their brand. [1011]There are regulatory burdens that these healthcare facilities have to report. So what are they regularly reporting to Health and Human Services that you would be able to have access to? What other types of, for facility-level surveys, that they might be commercial collected or publicly available datasets are there? [1005] Measurement-related themes and illustrative quotations …having locations to serve patient populations in different areas, patients who may or may not be able to travel to your main campus site… Many organizations have multiple satellite sites where they deliver care, whether the satellite sites are all located in wealthy neighborhoods or whether there’s a mix to deliver care in multiple different patient populations…that’s important to measure. [1007] … are you associated with the medical school would be black and white, or do you have a nursing school? That would be very knowable. … Who owns you? Are you owned by something? And then by some other entity, and when did that start? Very knowable. [1011] From a leadership perspective, you could always look at boards of directors of the organizations. You could always look at sort of C-suite leadership… The quality team might be a bit harder to get because it’s going to be lower down the org[anization] chart. And so, you might not have access to that info[rmation], but at least at the highest level, some of the high-level leadership, you would be able to get an understanding of diversity experience expertise within that. And I think that data could be collected pretty easily. [1012] Well, demographic characteristics generally are not that difficult to get. Payment characteristics are not too difficult to get, although it’s not easy. So, yeah. I wouldn’t say that they’re necessarily easy to collect, but there are good data sources. [1010] …there are some things potentially that could be measured indirectly, such as, using the navigator example, number of patients or percentage of your patients that are navigated or meet with a navigator on an annual basis. And that doesn’t necessarily directly give information about a navigation program, but it might indirectly. If it’s a low percentage, for example, it may not be a robust enough navigation program. [1004] …organizational culture is also really important, and it’s also difficult to collect … [service line] structure makes a difference. It is very hard to collect data on it because there are no commonly accepted terms. … One of my colleagues calls it a semantic swamp. [1006] …leadership factors, of course, are going to be probably the most difficult…very difficult to measure and to quantify leadership. [1010] …the soft touchy stuff, like how burnt out are people? How do people feel about the organization, satisfaction? Those kinds of things, I think, are hard because also we would lie to you in a survey. So I don’t even think it is worth doing…You won’t get a great representation. You’ll get people that are mad, people that love it and a lot of people who just aren’t going to fill [a survey] out or maybe they don’t want to complain about their organization. [1009] For cancer, the continuum is so important, …[the] operating linkages between different stages or organizations doing different work along the cancer continuum. … then operationally, how smoothly do they work together? Is information effectively being passed? Because, you know, at every point in those transitions there are substantial opportunities for fragmentation of care. [1006] We don’t know who we don’t take care of, necessarily, and so it’s a bit harder to measure whether we are providing equitable care. [1004] You could design surveys in particular ways that you could structure some of the answers that you want to get, that you would not be able to get from currently available data sources. … is this more of an exercise where we need to know what it is that we need to know and if it involves reaching out to individual systems and serving them to get an understanding of, for example, do they have some type of an equitable or care policy, or do they use race or social determinants as decisions in terms of to guide treatment or care? [1012] The other thing which I will say … will be hard is, they [the organization] may not want to give you all of this data … people are going to be really nervous about giving their financial stuff… people are going to be a little nervous about giving things that relate to power and structure. [1009] You could [not] get any data on performance evaluations, employee performance evaluations. You probably couldn’t get that kind of data. [1010] Well, the easiest to collect are things that are knowable, and organizations don’t feel could compromise their brand by sharing them. That the organizations would be comfortable associating with their brand. [1011] There are regulatory burdens that these healthcare facilities have to report. So what are they regularly reporting to Health and Human Services that you would be able to have access to? What other types of, for facility-level surveys, that they might be commercial collected or publicly available datasets are there? [1005] …historically, we always thought about leadership like senior leadership, which clearly is important. And we know that senior leadership, being onboard to do things, is often necessary, but not sufficient. We had seen for years that the role of middle managers was enormous in moving initiatives forward. [1002] …historically, we always thought about leadership like senior leadership, which clearly is important. And we know that senior leadership, being onboard to do things, is often necessary, but not sufficient. We had seen for years that the role of middle managers was enormous in moving initiatives forward. [1002] Values espoused by leaders were seen as shaping the organizational culture and determining what gets prioritized and thus monitored by the organization. Quality monitoring processes Respondents acknowledged the importance of routine monitoring to cancer care quality. Quality monitoring included not only how the organization routinely leveraged data to promote the delivery of evidence-based care processes, but, consistent with Donabedian’s model of quality assessment [12] also whether and how the organization monitored patient outcomes. For example, one respondent stated:I think a lot of times people think about process metrics, timeliness of visits and whatever process. But ultimately, I think process metrics are not the whole story for quality. It’s got to be related to patient outcomes. So, there’s got to be a way to actually assess patient outcomes as part of this organizational process and goals as well. [1007] I think a lot of times people think about process metrics, timeliness of visits and whatever process. But ultimately, I think process metrics are not the whole story for quality. It’s got to be related to patient outcomes. So, there’s got to be a way to actually assess patient outcomes as part of this organizational process and goals as well. [1007] Others further elaborated on this idea by highlighting the variety of patient-reported outcomes that could be considered:And then if you’re thinking [of] patient-reported outcomes, are these home symptom monitoring, are they other POS [point of service], distress assessments, geriatric assessment? And there’s lots of different things in that space that are very much infrastructure pieces at an organizational level. [1009] And then if you’re thinking [of] patient-reported outcomes, are these home symptom monitoring, are they other POS [point of service], distress assessments, geriatric assessment? And there’s lots of different things in that space that are very much infrastructure pieces at an organizational level. [1009] In addition, respondents noted that quality monitoring within an organization should be conducted to understand the performance of individuals and service units. Similarly, the ability of an organization to monitor care processes and outcomes within specific patient groups was seen as important. For example, one respondent questioned:Are these metrics specifically pulling out cancer information…, can you drill down to a particular diagnosis or a particular patient population…and are you breaking it out by vulnerable subgroups? [1005] Are these metrics specifically pulling out cancer information…, can you drill down to a particular diagnosis or a particular patient population…and are you breaking it out by vulnerable subgroups? [1005] In addition to what and how quality was monitored, the timeliness of feedback was seen as critical, as timely feedback is needed to enable identification of quality concerns and to support proactive interventions and course corrections. For example, a respondent stated: … you want to have sort of rapid feedback so that if you’re delivering care to a patient and somebody’s supposed to do something and they don’t, you can identify it quickly and follow up with that clinician or care team or whoever shortly after that visit so it’s a learning experience and it’s not like a report card that comes a month later where you said, you gave care to 100 patients this month, and oh, by the way, you messed up on 30 of them. [1001] … you want to have sort of rapid feedback so that if you’re delivering care to a patient and somebody’s supposed to do something and they don’t, you can identify it quickly and follow up with that clinician or care team or whoever shortly after that visit so it’s a learning experience and it’s not like a report card that comes a month later where you said, you gave care to 100 patients this month, and oh, by the way, you messed up on 30 of them. [1001] External incentives drive organization priorities External incentives, which create accountability for quality-related structure, processes, or outcomes, were perceived as important to determining organizational infrastructure, monitoring practices, and priorities. As one participant noted, asking “what’s your reporting relationship when you’re looking at quality of care metrics, what rating systems do you follow? What are you responding to? Do you break it out by product line? [1005] could be helpful. Likewise, another participant noted “…sometimes [improvement efforts] relate to incentives that are happening either internally within a system through leadership or nationally through reimbursement or Medicare or some of that accountability issues with hospitals.” [1008] Because of this, understanding the external accreditations and affiliations in which an organization participates was seen as helpful to being able to characterize an organization’s priorities and quality infrastructure. Likewise, simply knowing which of these external accreditations and affiliations an organization faced could further shed light on the types of quality performance monitoring the organization was doing as well as which external incentives are likely to drive areas of focus for improvement. For example, one participant noted: I think things like NCI [National Cancer Institute] designation status matters, so different accreditations, whether it’s Commission on Cancer, NCI, American Society of Clinical Oncology's Quality Oncology Practice Initiative (QOPI), thinking about those different levels. And there’s lots of different accreditations that are out there, but I think those do influence because they have a direct influence on the processes at the institution. They’re a higher- level view of, “Do they have programs in place to look at quality structure, navigation, treatment planning,” those types of things.[1009] I think things like NCI [National Cancer Institute] designation status matters, so different accreditations, whether it’s Commission on Cancer, NCI, American Society of Clinical Oncology's Quality Oncology Practice Initiative (QOPI), thinking about those different levels. And there’s lots of different accreditations that are out there, but I think those do influence because they have a direct influence on the processes at the institution. They’re a higher- level view of, “Do they have programs in place to look at quality structure, navigation, treatment planning,” those types of things.[1009] Support service access The provision of support services was reported as central to providing high quality cancer care. Support services include those things which facilitate a patient’s receipt of evidence-based and timely cancer treatment. Patient navigation services, psycho-oncology programs, nutrition services, patient instruction for medication use, and transportation-related assistance were all perceived as improving care quality. For example, it would be important to consider “Do you have a psycho-oncology program? If you have a psycho-oncology program, how does one get into that? Is it referral or is that automated? Do you have a nutrition program? [1009] On the other hand, voids in navigation were perceived as problematic to care quality. For example, a participant noted: Those patients who have not had the opportunity to be navigated through the system are potentially patients where something didn’t go the way it should go, or they had less access potentially to something, or delayed access to something. [1004]. Those patients who have not had the opportunity to be navigated through the system are potentially patients where something didn’t go the way it should go, or they had less access potentially to something, or delayed access to something. [1004]. Another participant suggested asking “if it’s something where people lack transportation to come into the clinic…do they offer sort of bus tokens or [ridesharing service] to the appointment or something like that, right? Where there’s some of those things that they can use to help make that people are getting into the clinic or do they have care coordinators … following up and with people answering questions. [1001] As such, examining available supportive services and programs, and how patients access such services, was thought to be central to understanding an organization’s cancer care quality. Similarly, another participant alluded to the complexity of cancer care and how that complexity across a broad care continuum warrants explicit efforts at care coordination:… For cancer, the continuum is so important, and what we didn’t address were actual operating linkages between different stages or organizations doing different work along the cancer continuum. So, you might look at whether there is a common ownership, for example, between a long-term care facility, a rehab[ilitation] facility, and an acute care hospital, just to mention three on the continuum. And that’s one piece of information, but then operationally, how smoothly do they work together? Is information effectively being passed? Because you know at every point in those transitions there are substantial opportunities for fragmentation of care. [1006] … For cancer, the continuum is so important, and what we didn’t address were actual operating linkages between different stages or organizations doing different work along the cancer continuum. So, you might look at whether there is a common ownership, for example, between a long-term care facility, a rehab[ilitation] facility, and an acute care hospital, just to mention three on the continuum. And that’s one piece of information, but then operationally, how smoothly do they work together? Is information effectively being passed? Because you know at every point in those transitions there are substantial opportunities for fragmentation of care. [1006] As such, an organization’s ownership, relationships, and affiliations were seen as one way to elucidate how entities within and across organizations work together and share information for care coordination. Extent to which approach to quality is holistic Respondents emphasized the importance of adopting a holistic approach to cancer care quality – prioritizing, monitoring, and adapting care delivery when gaps in quality are identified within and across organizations. For example, one participant stated …you have to have a process to identify the right metrics and then collecting accurate data and then a process to review those data and act on them. I think the whole process, I think, is necessary. [1007] These sentiments were echoed consistently each time with a focus on how important it was to not forget about actual improvement. Others went on to expand on this idea, focusing on an organization’s underlying capacity. For example, a participant stated. …we need to stop focusing on individual interventions and instead focusing on building capacities within organizations, like say, hospitals, to actually adopt any new intervention by having people who understand how to actually adopt, spread and scale the use of any intervention within that organization. [1002] …we need to stop focusing on individual interventions and instead focusing on building capacities within organizations, like say, hospitals, to actually adopt any new intervention by having people who understand how to actually adopt, spread and scale the use of any intervention within that organization. [1002] Likewise, participants highlighted how important continual learning was to quality: Do we know what the causes of those acute care events [are]… And then we have now, as a system, agreed that we’re going to do X, Y and Z to try and avoid those. And so, then, again, you’ve got your processes in place to try and avoid that. And then, if it turns out that there is an event, then you’re sort of looking at what happened after the fact to figure out where it broke down. [1001] Do we know what the causes of those acute care events [are]… And then we have now, as a system, agreed that we’re going to do X, Y and Z to try and avoid those. And so, then, again, you’ve got your processes in place to try and avoid that. And then, if it turns out that there is an event, then you’re sort of looking at what happened after the fact to figure out where it broke down. [1001] Continuous learning was seen as vital, as was the need for organizations to identify the causes of care-related events and implement processes to prevent them in the future. Organization-level features important to care equity Intentionality of an equity focus An intentional focus on equity was described as critical. For example, one participant noted how “Every design element that you have in an institution, if you are not having equity at the front of your mind, has the potential to increase disparities. And so, I think you have to be really purposeful in terms of putting in programs and procedures and thinking about how that’s going to impact equity. [1009] In addition to explicit commitments to equity in the mission and goals of the organization, people at all levels of the organization need to focus on equity and have taken concrete steps to ensure that equity is promoted throughout an organization. As noted by a participant one way to get to this intentionality is diversity in human resources:one of the best ways to ensure equity is to be intentional about it, and to start with an equitable opportunity across your organization, from clinicians, to leaders, to administrators, that a diverse enterprise is more likely to prioritize equity than a very uniform organization in terms of the people that make decisions. [1011] one of the best ways to ensure equity is to be intentional about it, and to start with an equitable opportunity across your organization, from clinicians, to leaders, to administrators, that a diverse enterprise is more likely to prioritize equity than a very uniform organization in terms of the people that make decisions. [1011] Drivers of care equity Respondents discussed the many current policies, practice models, reimbursement, historical practices, and other factors that drive current practices and how these multilayered factors also affect an organization’s focus on equity. Just as for care quality in general, respondents noted how external incentives, such as those coming from the National Cancer Institute (NCI) or Medicare, can impact if and what leadership prioritizes in terms of equity. For example, a participant notedthere’s a lot more incentives now, at the hospital level, in terms of reimbursement. I think that there are probably financial incentives long term. I think now, especially that we’re in a moment where the accountability issue in terms of being an organization that is anti-racist, that is actually serving all populations. This is something that now leaders are wanting to say that they’re doing, they want to show that. So, I think that there is more motivation and incentive at the organizational level than there probably has been historically. [1008] there’s a lot more incentives now, at the hospital level, in terms of reimbursement. I think that there are probably financial incentives long term. I think now, especially that we’re in a moment where the accountability issue in terms of being an organization that is anti-racist, that is actually serving all populations. This is something that now leaders are wanting to say that they’re doing, they want to show that. So, I think that there is more motivation and incentive at the organizational level than there probably has been historically. [1008] In addition to the impact of external incentives, success in delivering equitable care is affected by an organization’s infrastructure and resources, especially support services such as translation services, social work, and patient navigation. As one participant noted “While we have implemented the [electronic health record's] Social Determinants Module, there was a lot of pushbacks and continues to be from clinicians, with saying … if I identify these problems, what am I going to do about it? What resources do I have? I don’t have enough social workers. There’s not enough resources in the community.” [1005] Features of leadership are also critical to understanding how equity is addressed within an organization. These include whether the organization’s leadership prioritizes equity and whether there are designated “administrative roles that focus on delivering equitable care” [1011] within the leadership structure that have authority to support equity at the organizational level. Diversity among the individuals who serve in leadership roles is also influential, including whether they reflect the populations served by their healthcare system. In addition, respondents described the organization’s broader context as critical to achieving equity. For example, “to address [equity] you can’t just look at the treatment that’s provided within a particular institution. You’ve got to consider that in the broader community as a whole.” [1010] Data availability Explicitly determining key equity-related outcomes and ensuring that data to measure these outcomes are collected, analyzed, and reported (at least internally) were also seen as essential to supporting equity-related efforts. For example, a participant noted: “if we really don't attempt to look at quality of care by some of these characteristics like race, ethnicity, sex, in reality that we know that we don't really have health equity." [1005] Also endorsed was a critical need to understand the fit between available services and the population served. For example, a participant noted “We don’t know who we don't take care of, necessarily, and so it's a bit harder to measure whether we are providing equitable care.” [1004] As elaborated on by another participant, this concept of “fit” included geographic service availability as well as service needs: There’s no way to understand whether equity is an issue unless you have the patient-level demographic factors to be able to look at that data and then act on it. And so, there's got to be a way systematically for the organization to try to assess patient race, ethnicity, which are often issues related to equity, and also, in many regions of the country, whether [patients] live in urban or rural settings. Because I think distance to care and rural can also drive equity issues. So, there's got to be a way systematically for an organization to collect that kind of data. [1007] There’s no way to understand whether equity is an issue unless you have the patient-level demographic factors to be able to look at that data and then act on it. And so, there's got to be a way systematically for the organization to try to assess patient race, ethnicity, which are often issues related to equity, and also, in many regions of the country, whether [patients] live in urban or rural settings. Because I think distance to care and rural can also drive equity issues. So, there's got to be a way systematically for an organization to collect that kind of data. [1007] Measurement-related themes Some organization-level features are currently knowable Some organization-level features thought to be important to cancer care quality were described as currently feasible to compile given sufficient resources (Table 2). These include features available via public databases (e.g., the American Hospital Association’s annual survey) and compiled from information routinely available on health care organizations’ websites. For example, using data typically available on an organization’s website, it would be possible to characterize the geographic accessibility of the cancer care facilities owned and/or affiliated with an organization. Similarly, publicly available information might provide insights into an organization’s leadership structure and composition. Albeit not publicly available, data routinely compiled as a by-product of cancer care delivery, aggregated in innovative ways, could help determine other structural characteristics of an organization. For example, using the person-level characteristics available within electronic health records, some things about the socio-demographic composition of an organization’s patients could be determined. Such aggregated data could also provide insights regarding the availability of supportive care services that are important to cancer care quality. Some important organization-level features would present challenges Measuring other important organization-level features would present challenges stemming from a combination of factors, including a lack of agreed upon terminology and gaps in valid measures. Challenges will also arise because patient-centered assessment of cancer care quality means understanding the quality of care each person gets across the full cancer care spectrum, regardless of where they receive their care. As such, it would be important to consider transitions across facilities and among organizations to understand how coordinated such transitions were. Data innovations will require willingness and prioritization Capturing organization-level features will require partnership with those who generate and own the data. An organization’s legal ability, willingness, and comfort sharing information need to be considered as even if data are available these constraints may determine if the data would or could be shared publicly. Cancer care organizations already face multiple reporting burdens. Working collaboratively with these organizations further could afford opportunities to capitalize on what is already being reported. Because external incentives play a role in both what cancer care organization’s measure and prioritize as well as what researcher’s study, an organization like the NCI through its accreditation processes and funding initiatives is well positioned to move the field forward. Discussion Understanding how organization-level features influence both patient quality and equity outcomes as well as the implementation of interventions designed to improve patient quality and equity outcomes remains a challenge, in part because data describing these features are not readily available within the databases routinely used to study patient-level quality or equity. Thus, in the context of cancer care while it is well understood that patient-level quality and equity outcomes vary across facilities [1, 11, 14, 15], and that organization-level features impact the successful implementation of evidence-based interventions, it is not well understood which specific organization-level features contribute to those differences. Using in-depth interviews with diverse experts, we identified organization-level features perceived as important to patient-level quality and equity outcomes in the context of cancer care. Some, mostly structural, features were seen as eminently available and reliably measured via existing data sources. Other features, while more challenging to determine, were also considered significant and deemed worthy of development, investment, and compilation. This recognition stems from the crucial role that organizations’ internal characteristics and their external context is thought to play in both cancer care quality and equity [16, 17] and the implementation of evidence-based interventions [8]. In the realm of healthcare, including cancer care, the relationship between quality and equity is not just a conceptual overlap but a deeply interconnected reality [16]. Our findings highlight several organization-level features as pivotal determinants of both in the context of cancer care. Among these, leadership structure, organizational culture, and access to support services stood out as consistent influences. Research in healthcare quality and equity has repeatedly demonstrated that strong leadership committed to the principles of equity and quality is a linchpin for success [18]. The importance of leveraging leadership, particularly local opinion leaders, has similarly been highlighted in the context of implementation science [19]. A culture that prioritizes inclusivity and fairness is equally vital, as it sets the tone for all aspects of care delivery [20]. Access to support services, including patient navigation, plays an indispensable role in reducing disparities in care outcomes and thus enhancing care quality [19, 20]. As highlighted by these findings, longitudinal real-world studies that shed light on how organizational changes in leadership, culture, and support service provision impact patient-level quality and equity outcomes are needed. Furthermore, the importance of external incentives was also highlighted by our findings. As pointed out by participant responses, in the United States these can include those needed for accreditation from the Commission on Cancer, NCI or other professional organizations. They also include incentives driven by reimbursement models. In the United States one such reimbursement incentive of growing influence in the context of oncology care is the transition to value based care. Although not highlighted per se by the thematic analysis results, many components of the Centers for Medicare and Medicaid Services Enhancing Oncology Care Model (EOM) were mentioned by respondents as important organization-level features. These included care coordination through the use of navigation services and the integration of patient-reported measures in clinical care. While navigation services are relatively common in the context of oncology services, monitoring patient-reported symptoms and other outcomes remains less common. Understanding how such monitoring is integrated within routine care processes and how different strategies impact patient outcomes are important areas for future study. Monitoring care quality is intrinsically linked to both the pursuit of care equity and quality. Robust monitoring and evaluation mechanisms help identify gaps in evidence-based practice and disparities in care delivery, enabling proactive interventions to address them. Consistent with the Institute for Healthcare Improvement’s Whole System Quality [21], this interplay between organization-level features supporting care quality and equity underscores the need for a holistic approach to healthcare improvement and an intentional approach to care equity. Understanding how these features are interconnected, how they synergize and reinforce one another, is pivotal to our understanding of how context influences patient outcomes as well as successful intervention implementation in cancer care. In many ways our findings reaffirm the usefulness of existing quality frameworks such as those proposed by Donabedian over 30 years ago for measuring quality [12] or, more recently, advocated by the Institute for Healthcare Improvement for supporting care quality in practice [21]. They also highlight the importance of many of the domains considered within implementation science frameworks such as the Consolidated Framework for Implementation Science [8, 22] as well as the interconnectedness of an organization’s inner and outer setting. But they also point to opportunities to advance the field by considering additional granularity regarding how organization-level context is considered when care quality is measured and monitored in practice as well as the perceived importance of integrating patient-reported measures into the provision of care, and, at least in the context of cancer care, the availability of supportive services. As such findings here reinforce the need for frameworks not to be seen as static but as in need of continual evolution [22]. In addition, our findings highlight the need to develop organization-level measures that are feasible and scalable. It will be through such measurement advances that the data infrastructure provided by CDRNs, and emerging clinical data warehouses are enhanced, and we begin to understand the nuances that exist and need to be considered among different organizational contexts. For example, leadership structure and organizational culture arguably are seen as so influential to supporting patient outcomes, we can no longer accept our inability to characterize and improve them. Measuring organization-level features on scale has not been a priority to date in the United States, so it is not surprising that such capabilities do not yet exist, despite emerging capabilities elsewhere when data are not privately held. In the short term, efforts to address these challenges by leveraging publicly available data and collaborating with healthcare facilities hold promise for advancing our understanding of these critical factors. Our study is not without limitations. Perhaps foremost among these is that we did not design the study with a goal of achieving thematic saturation. Instead, we purposely selected a diverse group of experts to provide an initial step toward understanding how organization-level features support (or undermine) cancer care quality. Importantly, while some of the oncology care delivery experts directed NCI-designated cancer centers and/or held clinical leadership positions within oncology, omitted was the voice of oncology service administrative leaders. As such, important organization-level features that impact patient outcomes may not have been identified. Similarly, important measurement considerations may have been missed. Despite the richness afforded by the cancer care context, our interview guide did not explore which organization-level features were perceived as more or less important along different points or services in the cancer care continuum. As such whether different features are more or less important along the cancer care continuum was not explored and the extent to which our findings generalize to other clinical contexts is not known. The thematic results shed light on several aspects of organizational context that likely impact patient outcomes within healthcare organizations and offer practical steps for moving the field forward. These steps include: leveraging current publicly available information for studies evaluating the impact of organization features on patient-level care quality and equity,fostering partnerships between health care organizations and researchers to expand the types of organization-level features available within existing databases, and.creating opportunities and incentives for new measurement development and reporting as well as the joining of organization-level features within what have been rapidly evolving clinical data warehouses and repositories. leveraging current publicly available information for studies evaluating the impact of organization features on patient-level care quality and equity, leveraging current publicly available information for studies evaluating the impact of organization features on patient-level care quality and equity, fostering partnerships between health care organizations and researchers to expand the types of organization-level features available within existing databases, and. fostering partnerships between health care organizations and researchers to expand the types of organization-level features available within existing databases, and. creating opportunities and incentives for new measurement development and reporting as well as the joining of organization-level features within what have been rapidly evolving clinical data warehouses and repositories. creating opportunities and incentives for new measurement development and reporting as well as the joining of organization-level features within what have been rapidly evolving clinical data warehouses and repositories. Collectively, these findings begin to illuminate a path towards understanding how organization-level features support both cancer care quality and the implementation of evidence-based interventions, and thus improve patient outcomes and equitable care. We are hopeful that healthcare leaders, policymakers, and researchers will come together to advocate for the development of such measures as well as their inclusion within Clinical Data Research Networks (CDRNs) and other emerging databases. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1