Patient and Caregiver Considerations and Priorities When Selecting Hospitals for Complex Cancer Care

Abstract

Background:

Healthcare policies have focused on centralizing care to high-volume centers in an effort to optimize patient outcomes. However, little is known about patients’ and caregivers’ considerations and selection process when selecting hospitals for care. We aim to explore how patients and caregivers select hospitals for complex cancer care and to develop a taxonomy for their selection considerations.

Methods:

Qualitative study in which data were gathered from in-depth interviews conducted from March to November 2019 among patients with hepatopancreatobiliary cancers who were scheduled to undergo a pancreatectomy (n=20) at a metropolitan, urban regional or suburban medical center and their caregivers (n=10).

Results:

The interviews revealed 6 broad domains that characterized hospital selection considerations: hospital factors, team characteristics, travel distance to hospital, referral or recommendation, continuity of care, and insurance considerations. The identified domains were similar between participants seen at the metropolitan center and urban/suburban medical centers with the following exceptions: participants receiving care specifically at the metropolitan center noted operative volume and access to specific services like clinical trials in their hospital selection; participants receiving care at urban/suburban centers noted health insurance considerations and having access to existing medical records in their hospital selection.

Conclusions:

This study delineates the many considerations of patients and caregivers when selecting hospitals for complex cancer care. These identified domains should be incorporated into the development and implementation of centralization policies to help increase patient access to high quality cancer care that is consistent with their priorities and needs.

INTRODUCTION

In our current health care climate, there has been an increasing emphasis on outcomes improvement and cost containment. This paradigm has resulted in efforts to improve transparency for health care consumers, such as center of excellence accreditations^1–4 and publicly reported hospital ratings.⁵ This is especially important in cancer care, where there is significant outcome variability amongst hospitals,^{4, 6–8} with some hospitals being associated with a 4-fold higher risk of dying from complex cancer operations when compared to others.⁹

Consequentially, health systems are centralizing complex care by restricting low-volume hospitals from performing complex operations.¹⁰ These restrictions have raised concerns that such policies may compromise patient autonomy and access to care,^11–13 with studies demonstrating that vulnerable populations may be affected the most.^{14, 15} However, little is known about how patients actually select hospitals for their care. Before policies concerning receipt of care are broadly implemented, we first need to better understand patients’ and caregivers’ considerations and competing priorities in the matter.

The primary aim of this analysis was to explore in-depth how patients and caregivers select hospitals for complex cancer care. Secondarily, we assessed their impression of selected hospitals and providers which could have impacted their likelihood of staying with the facility, all of which could potentially inform the implementation of patient-centered health policies. We selected patients with hepatopancreatobiliary cancers who were scheduled to undergo pancreatectomy given the need for complex multidisciplinary care for these cancers and the prior literature suggesting the strongest associations between clinical volume and outcomes with the required operations.⁹

METHODS

We performed a qualitative study, based on semi-structured interviews conducted with patients with hepatopancreatobiliary cancer and their caregivers from March to November 2019. This study was approved by the Partners Healthcare (Protocol ID: 2018P002751) and University of Massachusetts’ (Protocol ID: H00017068) Institutional Review Boards. The manuscript was written in line with the Consolidated Criteria for Reporting Qualitative Research (COREG) checklist (supplemental material 1).

Participants

Eligible patients included adults with hepatopancreatobiliary cancer who were scheduled to undergo a pancreatectomy and were able to complete the interview in English. Eligible caregivers included family members or friends who accompanied the patient for the clinic visit. We used a purposive sampling approach to identify patients undergoing their procedures based on the location of medical centers where they were seen (metropolitan, urban regional and suburban) in the New England region. The metropolitan, urban regional and suburban medical centers included in this study perform approximately 200, 50, and 10 pancreatectomies per year, respectively.

Interview Guide

A semi-structured interview guide (supplemental material 2) was developed by our multidisciplinary team members, with expertise in medicine, health services, behavioral health and qualitative research methods (ZF, DC, LT) and with additional input from medical and surgical oncologists. The guide was designed to explore patient and caregiver perspectives on hospital selection for their pancreatectomy, including considerations, priorities, and extent of patient and caregiver involvement in the selection considerations. Open-ended questions and probes were used to elicit 1) factors that led patients to specific hospitals, and 2) impressions of the hospital and clinician after initial consultation and informed consent. The guide was revised iteratively following team discussions, and was cognitively tested with clinicians and patients to help enhance clarity and reduce bias in the questions.

Procedure

Clinic schedules at each hospital were screened by research staff a week ahead of time and confirmed with the patients’ surgical or medical oncologist. Identified eligible patients and caregivers were approached by study staff to discuss study participation at the end of the clinic encounter. Study staff conducted informed consent procedures and obtained written informed consent from interested patients. For patients who provided written informed consent, we also approached their caregiver for consent and participation. Consenting participants and caregivers completed a self-reported demographic form (supplemental material 3) and underwent semi-structured interviews with study staff (ZF, PL, RH). The interviewers are providers with training in qualitative methodology and had no established relationship with participants. Interviews lasted approximately 45 minutes and were conducted in-person or by telephone to reduce participant burden. Caregivers were interviewed separately from patients for those who agreed to independent interviews. Field notes were made during the interview, which were digitally audio-recorded and professionally transcribed. There were no repeat interviews, and transcripts were not returned to participants.

Data analysis

A content-analysis approach was used to evaluate the data.¹⁶ First, the responses were analyzed using a process of data immersion, and a coding framework was formed through an iterative process with the multidisciplinary team. Two research team members (ZF, PL) independently coded the first five patient and caregiver transcripts using NVivo 12 (QSR International, Burlington, MA). Flexibility was allowed to revise the coding structure as the reviews progressed. The team members reviewed coding consistency and met with a third team member (LT) to resolve discrepancies and revise the coding framework as needed. The revised framework was used to code the remaining transcripts. The research team consisted of health services researchers, oncologists, and a psychologist with expertise in public health, surgery, oncology, and qualitative research, which facilitated in-depth and clinically informed data interpretation.

The codes were organized and categorized into sub-domains and domains. Thematic saturation was determined to have been achieved when no new information was identified for three consecutive transcripts. Differences in domains between patients and caregivers, as well as participants seen at metropolitan and regional urban/suburban hospitals were explored.

RESULTS

Study participants

We conducted 25 interviews, including 20 individual interviews with patients (n=15) and caregivers (n=5), and 5 dyadic interviews with patient-caregiver pairs. Sixteen (64%) of the interviews were conducted in-person. Ten (40%) of the interviews were performed with patients and caregivers receiving care at a metropolitan hospital, whereas 15 (60%) were with participants seen at an urban regional or suburban hospital. Patient and caregiver characteristics are summarized in table 1.

Table 1.

Characteristics of participants stratified by patient and caregiver statuses.

Patients
Total, n		20
Urban/suburban medical centers		12 (60.0%)
Age, median (IQR)		70.5 (67.5, 73.0)
Female		8 (40.0%)
Hispanic origin		2 (10.0%)
Race	White or Caucasian	19 (95.0%)
Pathology	Pancreatic ductal adenocarcinoma	15 (75.0%)
	Cholangiocarcinoma	2 (10.0%)
	Pancreatic neuroendocrine tumor	2 (10.0%)
Travel time to primary care, median (IQR)		15.0 (10.0, 30.0)
Travel time to cancer facility, median (IQR)		40.0 (15.0, 90.0)
Education category	Less than high school	1 (5.0%)
	High school graduate (grade 12 or GED)	2 (10.0%)
	Some college or technical/vocational school	6 (30.0%)
	College graduate	4 (20.0%)
	Post-graduate or professional degree	7 (35.0%)
Employment status	Employed	6 (30.0%)
	Retired	13 (65.0%)
	Not employed for pay	1 (5.0%)
Household income	12,000 - 24,999	1 (5.0%)
	25,000 - 49,999	6 (30.0%)
	≥50,000	8 (40.0%)
	I would rather not disclose	5 (25.0%)
Payer category	Private	6 (30.0%)
	Medicaid	2 (10.0%)
	Medicare	12 (60.0%)
Marital status	Married or living with a partner	15 (75.0%)
	Widowed	2 (10.0%)
	Separated	3 (15.0%)
Children		18 (90.0%)
Living with children		5 (25.0%)
Living alone		2 (10.0%)
Caregivers
Total, n		10
Urban/suburban medical centers		5 (50%)
Age, median (IQR)		67.0 (58.0, 72.0)
Female		4 (40.0%)
Relationship to patient	Parent	1 (10.0%)
	Child	3 (30.0%)
	Spouse	6 (60.0%)
Hispanic origin		2 (20.0%)
Race	White or Caucasian	9 (90.0%)
Travel time to primary care, median (IQR)		15.0 (15.0, 35.0)
Travel time to cancer facility, median (IQR)		40.0 (15.0, 60.0)
Education category	Less than high school	1 (10.0%)
	High school graduate (grade 12 or GED)	1 (10.0%)
	Some college or technical/vocational school	5 (50.0%)
	Post-graduate or professional degree	3 (30.0%)
Employment status	Employed	5 (50.0%)
	Retired	5 (50.0%)
Household income	25,000 - 49,999	1 (10.0%)
	≥50,000	7 (70.0%)
	I would rather not disclose	2 (20.0%)
Payer category	Private	4 (40.0%)
	Medicaid	3 (30.0%)
	Medicare	3 (30.0%)
Marital status	Married or living with a partner	7 (70.0%)
	Widowed	2 (20.0%)
	Separated	1 (10.0%)
Children		9 (90.0%)
Living with spouse		7 (70.0%)
Living with children

Domains impacting hospital selection considerations

Factors impacting hospital selection were categorized into 6 broad domains: (1) hospital factors, (2) team characteristics, (3) travel distance from home to hospital, (4) referral or recommendation, (5) continuity of care, and (6) insurance considerations. These factors identified were themes that influenced the patient’s choice of the specific hospital that were knowable ahead of their first visits. Each domain with corresponding sub-domains and selected illustrative quotes are listed in table 2.

Table 2.

Taxonomy and illustrative quotes for domains identified in patients’ hospital selection considerations.

Domains	Sub-domains and illustrative quotes
	Reputation of hospital “I’ve gone on the internet over the years and looked up and it’s like the best hospitals in the United States for the best survival rates for certain kinds of cancer treatment, always in the top five hospitals are—[my hospital name].“ – patient.
	Availability of satellite centers close to home “If we do radiation and chemotherapy, I can do that here at [local satellite center] as well. It’s really the flexibility, and the timing, and the location of the satellite centers that made it a lot more advantageous to do the way I’m doing it as oppose coming downtown for everything.” – patient.
Hospital factors
	Favorable prior experience with hospital “I think the most important thing is the quality of the surgery that I received when I had my kidney taken out and the quality of the surgery that I had when I had my lung resection. The doctors were so good. The surgery was so good. I felt like I couldn’t have gotten better anywhere in the world.“ – patient.
	Reputation of surgeon “He’s an expert at what he does, and he’s written papers on that cancer. And plus, he’s a teaching professor, as my doctor is. Not only do they know what they’re doing, but they’re doing the latest and the greatest. And how can you beat that?” – patient.
Team characteristics	Operative volume “Because I read several times, ‘Don’t ever have a surgeon work on your pancreas unless he’s done 150 Whipple procedures [laughter].’ So they’re like, make sure. Find out beforehand that your surgeon has done that.“ – patient.
	Prior experience with other team members “Radiation machine was broken. I was not impressed with that, and I didn’t like the social worker there. I didn’t like the doctor, and I wouldn’t go there myself.” – patient.
	From an existing care clinician “My primary care physician… recommended him. She said the local community hospital, [hospital’s name] the surgical team they have here is good.” – patient.
Referral or recommendation	From a new care clinician “I believe [surgeon] was a referral from maybe one of the clinicians at [hospital’s name] probably. I was in there for an [endoscopic retrograde cholangiopancreatography].” – patient.
	From a non-clinician “My daughter has a very good friend from like 40 years ago and she happens to work for a lobbyist for a radiation oncologist, and she has a really good contact in the radiology department at [hospital’s name]. And she emailed him and he emailed back and said, ‘Anybody at the [center’s name] is excellent and you won’t go wrong no matter who you choose.’” – patient.
Travel distance from home to hospital	Prioritizing hospital proximity due to pragmatic concerns “It’s not as bad as going to [city name]. I mean that was part of why I didn’t want to go to [city name]. It’s a long way. And then coming out of surgery and you’ve got that drive home. And two hours on the [free way], with that kind of surgery, so-“ – patient.
	“We have… a small group of friends. They’re all going to want to come and visit. If [patient is] at [more distant hospital], that’s not going to happen. I’ve got to work every day. You know what it’s like. I don’t have a choice.” – caregiver.
	“Oh, yeah. I mean it takes me four hours to get to [city name]. No, I don’t know anybody there. But then I just stay here. Yeah, I have an apartment. I have family. My grandchildren live 15-minute walk away and they come by. There’s nothing like having family when you’re going through something.” – patient.
	Prioritizing perceived hospital quality over proximity “Driving’s not a problem. If not, let the ambulance take me. I don’t care. I’d rather go to a better facility than a worse facility. In other words, if there’s something better out there, I’d like that. I only got one life to live. And I don’t want to give it up too easily.” – patient.
Continuity of care	Access to existing medical records “They have access to my records right away. If they need to look at an EKG from six months ago, they have it. They don’t have to call somebody and say fax it to me. I did that because I figured every bit of information, whatever doctor, I think, has is just going to benefit me.” – patient.
Insurance considerations	Insurance approval “I have [insurance name], so in other words, when I was going to go to [hospital’s name], they didn’t want to do anything about anything until you show them [CT] scans. So I’ve gotten a [CT] scan at [hospital’s name] and by the time I got accepted to get the [CT] scan my appointment was canceled at [hospital’s name] and it took a kind of rollercoaster backwards.” – patient.

Domain 1a: Hospital factors

One of the main hospital factors that influenced hospital selection was perceived hospital reputation. For instance, participants commonly perceived that a hospital’s affiliation with a university or medical school implied that the hospital’s equipment would be state-of-the-art. Participants also favored hospitals that have satellite centers closer to home, which allowed them to receive chemotherapy locally. Lastly, favorable prior experiences with treatments or procedures at a hospital drove patients back to the same hospital for their cancer care.

Domain 1b: Team characteristics

Participants commonly identified certain team characteristics that drove them to seek care at selected hospitals. For instance, several participants identified operative volume as a metric for quality care, whereas others identified the surgeon’s publication record to indicate surgeon expertise. Participants’ prior positive or negative experiences with other non-oncology team members (i.e. social workers) also impacted their decisions to pursue care at selected hospitals.

Domain 1c: Travel distance from home to hospital

Participants varied in whether they were concerned about travel distance from home to hospital. Participants who considered travel distance in hospital selection wanted to make things more familiar and easier for family members to provide transportation and/or on-site emotional support during surgery, inpatient recovery, and follow up visits. Participants also identified the burdens of paying for parking and/or local housing for extended periods of time when undergoing an operation and follow-up care at a more distant location. In comparison, other participants were willing to bargain travel distance for perceived hospital quality.

Domain 1d: Referral or recommendation

Referrals and recommendations played an important role in determining where participants received their cancer care. Participants commonly did not have specific clinicians in mind but relied on and trusted the referrals from their primary care physician or other clinicians (i.e., gastroenterology). Additionally, some participants selected hospitals based on recommendations from family and friends who knew clinicians at specific hospitals.

Domain 1e: Continuity of care

One patient emphasized continuity of care when explaining their hospital selection. Specifically, this participant felt that the hospital they attended previously would have direct access to existing medical records and prior studies performed without the need to request them from another hospital, and that every bit of information would benefit the oncology team.

Domain 1f: Insurance considerations

Lastly, some participants noted that health insurance parameters impacted hospital selection. One patient intended to seek cancer care at a specific hospital, but that hospital wouldn’t accept their insurance until cross-sectional imaging was performed, which could have taken a few weeks. By then, this participant had found another hospital that had accepted their insurance, and the promptness of being seen in the setting of a new cancer diagnosis was important to them.

Differences in domains and sub-domains between groups

There were no differences in domains between patient and caregiver dyads. Similarly, domains were consistent among participants seen at the metropolitan, urban regional, and suburban medical centers, with some exceptions. For example, participants receiving care at the metropolitan high-volume referral center noted operative volume and access to specific services like clinical trials in their hospital selection. Participants receiving care at the urban regional or suburban medical center noted health insurance considerations and having access to existing medical records in their hospital selection.

Domains identified from participants’ impression of hospitals and clinicians

Participants’ impression of hospitals and clinicians were categorized into five broad domains: (1) hospitality, (2) patient-clinician communication and rapport, (3) perception of expertise, (4) access to specific services, and (5) ability to navigate logistics of treatment plan. These impressions represent themes identified after their initial experiences that may have influenced their satisfaction and likelihood of staying with their chosen facilities. Each domain with corresponding sub-domains and selected illustrative quotes are listed in table 3.

Table 3.

Taxonomy and Illustrative quotes for domains identified from participants’ impression of hospitals and clinicians.

Domains	Sub-domains and illustrative quotes
Hospitality	Amenities “..you can always get a drink 24 hours a day over at [café name]. And I know that from other hospitals when he’s been there, there’s nothing really available.” – caregiver. Familiarity “I knew everybody. I can make a phone call and find out anything just like that.” – caregiver.
Patient-clinician communication and rapport	Understanding of treatment plan “He laid it all out. He gave us a lot of confidence on the path that we have to go through.” – caregiver. Ability to communicate effectively “The team was on top of things. Communicated with us as a family. We had great communication.” – caregiver.
Perception of expertise	Perceived quality of surgeon “He’s head of Surgical Oncology, and he’s a vice-president of Surgery in the hospital so his credentialing is excellent.” – patient. “If you google [surgeon’s name] you see a lot of articles about him, that he’s written in science journals. You see all his degrees, how long he went to [University’s name], where he did his residency,,” – patient. External appearance “It’s a beautiful facility with a terrific staff at the desk. Just coming into the building, one could draw the conclusion that [hospital’s name] is top-of-the-line.” – patient.
Access to specific services	Clinical trial enrolment opportunities “[Medical oncologist’s name] was the one that introduced us to the clinical trial, and that’s how we then came here. It was very upbeat, very, very encouraging.” – caregiver.
Practical access to care	Appointments “[Hospital name] called me back in the morning after I left my message and said “if you can come up here on this date, we have an appointment for you with a surgeon and an oncologist.” I mean, they were so responsive and on the ball that I went up.” – patient. Travel “Yeah, (the driving) too much. I basically jumped into saying “yeah yeah yeah” before you actually did it and went through it.” – caregiver.

Domain 2a: Hospitality

Several participants emphasized a favorable impression of hospitality. This included hospital services and gestures such as 24-hour access to hospital food services or being offered warm blankets during chemotherapy infusions. Participants also noted that parking reimbursement and valet services meant a lot to them, as parking should not be one of their worries in the setting of a new cancer diagnosis.

Domain 2b: Patient-clinician communication and rapport

Participants also emphasized the value of effective patient-clinician communication to their sense of security and confidence, such as the ability of their clinicians to explain the complex treatment plan and operations in lay-terms, and to keep both patients and their family caregivers well-informed. In comparison, being overwhelmed with too much technical information led participants to feel frustrated, insecure, and mistrustful of their clinical team.

Domain 2c: Perception of expertise

A prominent factor affecting participant impressions of their cancer care was the perceived quality of the surgical oncologist, related to the surgeons’ education, training, academic productivity, and professional titles. At the hospital-level, the external appearance (i.e. lobby appearance) of the hospital was a surrogate for quality of care for some participants.

Domain 2d: Access to specific services

An important factor identified that could’ve influenced patients staying with their facility of choice was access to specific services such as the availability of a clinical trial. Participants expressed that this provided a great sense of needed hope during a time when they were delivered a new diagnosis of cancer with a grim prognosis.

Domain 2e: Practical access to care

Lastly, participants reflected on their access to care and ability to follow through with the treatment plan after hospital selection. For instance, one patient and caregiver dyad transferred their cancer care to a facility closer to home after they realized they had underestimated the travel burden to and from the hospital. Participants also commonly emphasized that promptness and ease of scheduling appointments across the different specialties (medical, radiation, and surgical oncology) after the initial telephone encounter was important to them.

Differences in domains and sub-domains between groups

There were no differences in domains between patients and caregivers or between participants seen at the metropolitan, urban regional, or suburban medical centers.

Domains identified from patients who received opinions at more than one facility or transferred care

There were three identified patients who were seen at more than one facility, or had their care transferred to a different one after their initial experience. For the two patients who had multiple opinions, the main drivers of choosing one facility over the other were familiarity with the facility, proximity to home, and insurance considerations (expedited approvals for timely cancer care). For the one who transferred care, both the patient and caregiver realized the travel distance required to reach the metropolitan center was overwhelming post-hoc, and preferred a suburban facility closer to home. It is important to note that the study was not powered for this sub-analysis, and that thematic saturation is likely to have not achieved.

DISCUSSION

Recent efforts have grown to centralize complex cancer operations to high-volume centers to optimize clinical outcomes, which could result in limiting other medical centers from providing such services. However, little is known about how patients and their caregivers make decisions regarding hospital selection, with concerns that such efforts or potential policies may compromise their autonomy and access to care. Our taxonomy suggests that there are various reasons extending beyond clinical outcomes that drive patients to receive their care at specific hospitals, such as travel distance to the hospital, referrals and/or recommendations, continuity of care, and insurance considerations. We identified relative consistency in domains identified by patients receiving their care at metropolitan and urban/suburban medical centers.

Prior studies have attempted to delineate patients’ decision-making when it comes to selecting hospital for complex cancer care. Ejaz and colleagues surveyed patients seen in a tertiary referral center’s surgical clinic and reported that surgeon experience and specialized training were rated as the most important factors patients considered when choosing a cancer surgeon.¹⁷ However, surveyed patients were self-selected, given that they had already made the decision to travel to a high-volume referral center. Recently, Resio and colleagues sampled a nationally representative cohort and demonstrated that most were motivated to travel for improved outcomes; financial considerations (i.e. cost of transportation, parking and hotel) were the most common barrier and over 90% would travel if these barriers were removed.¹⁸ However, these were hypothetical scenarios and responses provided by a healthy cohort, which may not accurately reflect real-life decision-making. Additionally, the survey approach of this prior study limited responses only to domains identified by the investigators. In our current study, other important domains identified by patients with hepatopancreatobiliary cancers and their caregivers (i.e. trust in referral, comfort of being close to family, familiarity with the local hospital) suggest that successful centralization of complex cancer care is not as simplistic as it may seem.

Referrals represent a key factor influencing hospital selection. In our study, patients and caregivers viewed referrals as a prescription, and often placed a great deal of trust on them based on previously built relationships. Similar to what has been demonstrated in the primary care literature, current results suggest that patients were exercising “passive choice activation,” meaning that they remained with the default-referred provider without meaningfully considering alternatives.¹⁹ Moreover, while patients and caregivers may have conducted Internet searches during hospital selection, they commonly did not access publicly available performance information.¹⁹ These results build upon prior literature demonstrating that patients may have difficulty understanding the performance information provided to them, and they may lack trust in the underlying data.^{2, 20, 21} Therefore, they may instead rely on advice from family and friends^22–24 or from their referring primary care physician, with the latter being regarded as the most trustworthy source of information.^25–27 Consequently, efforts to increase transparency via public reporting of outcomes and accreditation processes may be optimized by engaging primary care physicians in these endeavors.

Patients and caregivers also highlighted the importance of prompt scheduling in the setting of a new cancer diagnosis, which was evident in both the initial hospital selection and impression of the initial clinical encounter. This sense of urgency is warranted, with studies suggesting that delays in treatment are associated with worse oncologic outcomes in other cancers.^28–30 In hepatopancreatobiliary cancers, there has been a paradigm shift towards upfront systemic treatment prior to surgical resection,^{31, 32} and the link between time to chemotherapy initiation and survival is unknown. More work is needed to understand whether removing barriers to prompt initiation of treatment may improve patient satisfaction, wellbeing, and potentially clinical outcomes. Our results also suggest the importance of examining the impact of centralization policies on delays in treatment initiation.

Some of the main concerns with centralizing complex cancer care were the potential impact on patient autonomy and spatial access to care.^{12, 14} Our study provides relevant insight into patient and caregiver priorities and concerns. Patients and caregivers who preferred to minimize travel distance identified that they wanted to be close to family during cancer treatment and/or to be treated at a hospital that was familiar to them. Such preferences often superseded the importance of clinical outcomes. These reasons may help to elucidate recent findings that many patients are reluctant to travel for better care,³³ with some even bypassing high-volume hospitals to receive their operations at low-volume hospitals.³⁴ Our results show the value of incorporating patient perspectives in the development and implementation of centralization policies to help increase patient access to high quality care that is consistent with their priorities and needs. Additionally, health systems could focus on within-system regionalization,³⁵ or regional quality collaboratives and work to establish remote multidisciplinary tumor boards to attenuate the volume-outcome relationship.^{36, 37}

This study should be interpreted in the context of important limitations. First, the study was conducted among a limited set of hospitals in Massachusetts, and may not be generalizable to other settings. Second, participants were predominantly white and were all English speaking, and there may be cultural differences in health beliefs that were not captured in our analysis.^{38, 39} An important next-step is to purposefully delineate such priorities and preferences in demographic and socioeconomic minorities to ensure that their needs are not overlooked when setting surgical policy. Nevertheless, we believe our findings still places a spotlight on the often overlooked importance of considering patient preferences in surgical policy.

In summary, this study delineates the many and multifaceted considerations of patients and caregivers when selecting a hospital for complex cancer care. Our findings can help to inform future research and initiatives related to health care centralization. This work identifies (1) actionable barriers and facilitators to care for patients and caregivers, (2) professional behaviors and practices that can improve patient-centered care for providers, and (3) actionable structural factors that can facilitate delivery of care for hospitals and organizations. Future studies can build upon our findings to test the relative extent to which each domain influences hospital selection, and to inform evidence-based, systems-level interventions to improve our delivery of patient-centered care.