Time to Transfer as a Quality Improvement Imperative: Implications of a Hub-and-Spoke Health System Model on the Timing of Emergency Procedures

Abstract

Background:

In the increasingly prevalent hub-and-spoke health system model, specialized services are centralized at a hub hospital, while spoke hospitals offer more limited services and transfer patients to the hub as needed. In one urban, academic health system, a community hospital without procedural capabilities was recently incorporated as a spoke. The goal of this study was to assess the timeliness of emergent procedures for patients presenting to the spoke hospital under this model.

Methods:

The authors performed a retrospective cohort study of patients transferred from the spoke hospital to the hub hospital for emergency procedures after the health system restructuring (April 2021–October 2022). The primary outcome was the proportion of patients who arrived within their goal transfer time. Secondary outcomes were time from transfer request to procedure start and whether procedure start occurred within guideline-recommended treatment time frames for ST-elevation myocardial infarction (STEMI), necrotizing soft tissue infection (NSTI), and acute limb ischemia (ALI).

Results:

A total of 335 patients were transferred for emergency procedural intervention during the study period, most commonly for interventional cardiology (23.9%), endoscopy or colonoscopy (11.0%), or bone or soft tissue debridement (10.7%). Overall, 65.7% of patients were transferred within the goal time. 23.5% of patients with STEMI met goal door-to-balloon time, and more patients with NSTI (55.6%) and ALI (100%) underwent intervention within the guideline-recommended time frame.

Conclusion:

A hub-and-spoke health system model can provide access to specialized procedures in a high-volume, resource-rich setting. However, ongoing performance improvement is required to ensure that patients with emergency conditions receive timely intervention.

Health system mergers provide an opportunity to restructure organizations and maximize their ability to provide high-quality care. Traditionally used in the transportation industry, the hub-and-spoke model of mergers focuses on strategic centralization of advanced medical resources (hub) with secondary entities (spokes) that offer more limited services.^1,2 The hub offers complex medical interventions that are labor- and capital-intensive and serves as the accepting facility for patients at spoke hospitals who require these services.^1,3 Ultimately, the goal of this model is to maximize market coverage while conserving resources.³ Even prior to the financial pressures experienced by health systems in the wake of the COVID-19 pandemic, the hub- and-spoke model has gained increasing attention due to its potential to reduce operating costs for participating hospitals.^1,4–6

Hub-and-spoke health system restructuring has the potential to influence outcomes for time-sensitive medical conditions. For small- and medium-sized hospitals, having a relationship with a tertiary referral center can optimize patient outcomes by providing access to specialists when necessary.^7–9 However, the process of transferring patients from a spoke to the hub may lead to delays in definitive care. This presents a dilemma for conditions in which time to treatment is critical for optimal clinical outcomes, such as myocardial infarction, limb ischemia, and necrotizing soft tissue infection. In two recent studies of patients with emergency general surgery conditions, transferred patients were more likely to experience in-hospital mortality and complications.^10,11 Thus, for patients with emergency procedural needs, the hub-and-spoke system must balance the competing priorities of access to expertise and timeliness of intervention.

This study analyzes a unique hub-and-spoke restructuring within a large academic health system and its implications for patients requiring emergency procedures. Our health system assumed management of a nearby safety-net hospital and established a hub-and-spoke relationship in which all patients requiring procedures that cannot be performed at the bedside are transferred to our tertiary referral center for treatment. As a first step toward quantifying the effect of this model on patient outcomes, the goal of our present study is to describe the timeliness of the transfer process for patients requiring emergency procedures in the first 1.5 years under this new arrangement.

METHODS

Setting

In the spring of 2021, our health system assumed management of a financially distressed hospital in our city, Hospital X. Hospital X provides care for a historically underserved patient population and handles approximately 45,000 encounters per year in its emergency department (ED). As our tertiary referral center, Hospital Y, is located 2.3 miles away, patients at Hospital X requiring any procedure that cannot be performed at the bedside are now transferred to Hospital Y for treatment.

Transfer Process

According to protocol, a transfer request is initiated from Hospital X via phone call to the central health system transfer center by the emergency medicine physician or hospitalist caring for the patient, as applicable. The transfer center then connects to the on-call clinician for the requested service at Hospital Y. In the rare circumstance in which Hospital Y does not provide the service that the patient requires (for example, orthopedic trauma surgery), the transfer is routed to the appropriate on-call clinician at one of our health system’s other two hospitals in the city. The on-call clinician accepts the transfer and chooses a transfer priority based on the details of the case. The transfer priorities are defined by our health system as follows: First priority confers a goal transfer time of ≤ 2 hours, second priority confers a goal transfer time of ≤ 8 hours, and third priority confers a goal transfer time of ≤ 24 hours.

For most cases, the transfer center must identify an available bed at Hospital Y to which the patient can be taken on arrival. Exceptions can be made if (1) no bed is available and the patient is expected to decompensate without immediate intervention, in which case an ED-to-ED transfer may be utilized; or (2) the patient requires cardiac catheterization, in which case the patient can be taken directly to the catheterization laboratory.

To arrange transportation, the transfer center contacts the dispatcher for the health system’s internal transportation service. The dispatcher determines whether the transfer will be completed by one of the health system’s ambulances or by an ambulance owned by a private company that contracts with the health system, depending on availability and level of care required.

Data Source and Study Cohort

The health system’s transfer log, a custom database that pulls information from transfer center documentation as well as the electronic medical record, was queried to identify all patients transferred from Hospital X to Hospital Y for emergency procedures between April 2021 and October 2022. Patient demographics and transfer details, including indication, priority, request time, and arrival time, were abstracted from the transfer log. Information about each patient’s first procedure, length of stay, and discharge were obtained via chart review.

Patients were included in the analysis if they underwent an emergency procedure after transfer from Hospital X to Hospital Y. An emergency procedure was defined as the first non-bedside procedure that occurred ≤ 72 hours after arrival to Hospital Y, performed by any surgical service, Interventional Cardiology, Gastroenterology, or Interventional Radiology, as identified by chart review. Patients may have been located in the ED or inpatient ward prior to transfer initiation.

Outcomes

Our primary outcome was the proportion of cases that arrived within the goal transfer time according to transfer priority. Each patient’s transfer time was defined as the time from the placement of the transfer request, as documented by the transfer center during the transfer call, to the time of patient arrival at Hospital Y.

Secondary outcomes included (1) time from transfer request to arrival at Hospital Y, (2) time from transfer request to procedure start, and (3) time from initial presentation, defined as arrival at the ED at Hospital X, to procedure start. To assess factors contributing to transfer delays, total transfer time was further divided into (1) time from transfer request to bed assignment, as an indicator of bed availability, and (2) time from bed assignment to arrival at Hospital Y, as an indicator of transport availability.

Subgroup Analyses

Given the heterogenous nature of emergency conditions that may require procedures, we also conducted subgroup analyses for three acute conditions for which clear evidence of the importance of timely intervention has been described and with which > 1 patient presented during the study period: acute ST-elevation myocardial infarction (STEMI), acute limb ischemia (ALI), and necrotizing soft tissue infection (NSTI). For STEMI, US guidelines recommend time from presentation to percutaneous coronary intervention of less than 90 minutes.¹² For ALI, society guidelines recommend time from presentation to revascularization within 6 to 24 hours, depending on severity.¹³ For NSTI, debridement within 12 hours of suspected diagnosis is recommended.¹⁴

Statistical Analysis

Descriptive statistics are reported as frequencies and percentages for categorical variables and as means with standard deviations or medians with interquartile ranges (IQRs) for continuous variables. Transfer times were compared between groups using Kruskal-Wallis tests. Statistical analyses were performed using Stata 17.0 (Stata Corp LLC, College Station, Texas).

This study was deemed to be exempt by the University of Pennsylvania Institutional Review Board due to its classification as a quality improvement initiative.

RESULTS

Cohort Characteristics

A total of 335 patients were transferred from Hospital X to Hospital Y for emergency procedures during the study period. The mean age was 56.3 (±16.9) years, and 53.4% of the cohort was female. The majority of patients (84.8%) identified as non-Hispanic Black. The majority (63.0%) of patients were transferred to surgical services, while 37.0% were transferred to medical services. The median length of stay including Hospitals X and Y was 6 days (IQR 3–11). The majority of patients (73.1%) were discharged home, while 3.9% died in-hospital. These demographic and clinical characteristics are summarized in Table 1. The most common procedure types patients required were interventional cardiology (23.9%), endoscopy or colonoscopy (11.0%), and bone or soft tissue debridement (10.7%) (Table 2).

Table 1.

Demographic and Clinical Characteristics of Patients Transferred from Hospital X to Hospital Y for Emergency Procedures (April 2021–October 2022)

Characteristic
Age in Years, mean (SD)	56.3 (16.9)
Gender, n (%)
Male	156 (46.6)
Female	179 (53.4)
Race/Ethnicity, n (%)
Non-Hispanic white	19 (5.7)
Non-Hispanic Black	284 (84.8)
Hispanic	12 (3.6)
Other or unknown	20 (6.0)
Primary Language
English	316 (95.2)
Non-English	16 (4.8)
Receiving Service, n (%)
Surgical	211 (63.0)
Medical	124 (37.0)
Days from Initial Presentation to Discharge, median (IQR)	6 (3–11)
Days from Transfer to Discharge, median (IQR)	5 (2–10)
Disposition, n (%)
Home	245 (73.1)
Acute rehabilitation facility	13 (3.9)
Skilled nursing facility	49 (14.6)
Left against medical advice	8 (2.4)
Deceased	13 (3.9)
Other	7 (2.1)

SD, standard deviation; IQR, interquartile range.

Table 2.

Distribution of Emergency Procedures Performed for Patients Transferred from Hospital X to Hospital Y (April 2021–October 2022)

Procedure Type	n (%)
Interventional cardiology procedure	80 (23.9)
Endoscopy/colonoscopy	37 (11.0)
Bone/soft tissue debridement	36 (10.7)
Cholecystectomy	35 (10.4)
Appendectomy	25 (7.5)
Interventional radiology procedure	20 (6.0)
Urology procedure	13 (3.9)
Lysis of adhesions	11 (3.3)
Brain/skull procedure	9 (2.7)
Vascular procedure	9 (2.7)
Bronchoscopy, tracheostomy, other otolaryngology procedure	8 (2.4)
Abdominal wall hernia repair	7 (2.1)
Rectum/anus procedure	7 (2.1)
Small bowel resection	7 (2.1)
Spine procedure	7 (2.1)
Cardiothoracic procedure	6 (1.8)
Stomach procedure	6 (1.8)
Gynecology procedure	5 (1.5)
Other	4 (1.2)
Colectomy/colostomy	2 (0.6)
Liver/pancreas procedure	1 (0.3)

Transfer Outcomes

Most patients were assigned either first priority (37.3%) or second priority (56.4%). Of the first priority cases, 57.6% arrived at Hospital Y within the goal time of ≤ 2 hours, with a median transfer time of 1.8 hours (IQR 1.3–2.8). Of the second priority cases, 73.0% arrived at Hospital Y within the goal time of ≤ 8 hours, with a median transfer time of 4.4 hours (IQR 2.6–8.1). The median time from transfer request to procedure start was 5.7 hours (IQR 2.7–19.7) for first priority cases and 22.9 hours (IQR 9.3–41.0) for second priority cases. Transfer outcomes by priority are summarized in Table 3.

Table 3.

Transfer Characteristics for Patients Transferred from Hospital X to Hospital Y for Emergency Procedures, by Transfer Priority (April 2021–October 2022)

Priority	# of Cases n (%)	Arrived Within Goal Time n (%)	Transfer Requests→Arrival Median hours (IQR)	Transfer Requests→Procedure Start Median hours (IQR)
First (≤ 2 hours)	125 (37.3)	72 (57.6)	1.8 (1.3–2.8)	5.7 (2.7–19.7)
Second (≤ 8 hours)	189 (56.4)	138 (73.0)	4.4 (2.6–8.1)	22.9 (9.3–41.0)
Third (≤ 24 hours)	21 (6.3)	10 (47.6)	26.0 (6.1–66.1)	42.5 (7.7–78.6)

IQR, interquartile range.

There was substantial variation in the time to bed assignment at Hospital Y as well as in transport availability. For all cases, the median time from transfer request to bed assignment was 0.8 hours (IQR 0.2–3.9), while the median time between bed assignment and arrival at Hospital Y was 2.1 hours (IQR 1.6–3.3). For cases that arrived within their goal time, the median time to bed assignment was 0.5 hours (IQR 0.1–1.6); for cases that were delayed, time to bed assignment was significantly longer at 5.3 hours (IQR 0.4–17.3, p < 0.001). Similarly, for on-time transfers, the median time from bed assignment to arrival was 1.8 hours (IQR 1.3–2.6), while for delayed cases, the median time from bed assignment to arrival was 2.8 hours (IQR 2.0– 4.9, p < 0.001).

The volume and timeliness of transfers varied month to month during the study period. The mean number of transfers per month was 18 (±4.6), and the mean proportion of patients who arrived within the goal time each month was 65.5% (±10.7%). Notably, the two months with the lowest proportion of on-time transfers coincide with two periods of peak bed flow challenges at Hospital Y. In November 2021 Hospital Y opened a new patient building, which required substantial training and acclimation for staff, and in February 2022 Hospital Y experienced capacity strain due to hospitalizations from the COVID-19 Omicron variant. Transfer patterns over time are depicted in Figure 1.

Figure 1:

This bar graph shows monthly volume of transfers from Hospital X to Hospital Y for emergency procedures and the proportion that arrived within goal time based on transfer priority (April 2021–October 2022). November 2021: Hospital Y’s new patient building opened; February 2022: capacity strain at Hospital Y due to Omicron surge.

Subgroup Analyses

All patients who presented to Hospital X with STEMI, ALI, and NSTI were transferred to Hospital Y as first priority. For the STEMI group (n = 17), 94.1% arrived at Hospital Y within the goal of ≤ 2 hours, with a median transfer time of 0.8 hours (IQR 0.7–1.2). For the ALI group (n = 4), 25.0% arrived on time, with a median transfer time of 2.2 hours (IQR 1.8–3.6). For the NSTI group (n = 9), 66.7% arrived on time, with a median transfer time of 1.7 hours (IQR 1.4–2.1). Compared to other first priority transfers, the median transfer time was significantly shorter for the patients with STEMI (p < 0.001) and NSTI (p = 0.035) but no different for those with ALI (p = 0.418). The transfer outcomes for these conditions are summarized in Table 4.

Table 4.

Transfer Characteristics for Patients Transferred from Hospital X to Hospital Y for Emergency Procedures, by Selected Conditions (April 2021–October 2022)

Condition	Cases n (%)	Arrived Within Goal Time n (%)	Transfer Requests→Arrival Median hours (IQR)	Arrival → Procedure Start Median hours (IQR)	Transfer Request→Procedure Start Median hours (IQR)	Presentations→Procedure Start Median hours (IQR)
STEMI	17 (5.1)	16 (94.1)	0.8 (0.7–1.2)	0.2 (0.1–1.1)	1.2 (0.9–2.1)	2.8 (1.6–3.5)
ALI	4 (1.2)	1 (25.0)	2.2 (1.8–3.6)	2.8 (2.0–6.1)	4.7 (4.1–9.4)	8.7 (5.7–15.9)
NSTI	9 (2.7)	6 (66.7)	1.7 (1.4—2.1)	3.1 (2.8–4.8)	5.7 (4.3–6.1)	10.1 (5.6–17.5)

IQR, interquartile range; STEMI, ST-elevation myocardial infarction; ALI, acute limb ischemia; NSTI, necrotizing soft tissue infection.

For these conditions, the proportion of patients whose procedures started within the guideline-recommended time frame varied. For STEMI cases, median time from presentation at Hospital X to procedure start at Hospital Y was 2.8 hours (IQR 1.6–3.5), with only 23.5% of cases meeting the door-to-balloon guideline of less than 90 minutes. For patients with ALI, in whom goal time to intervention is 6 to 24 hours depending on severity, the median time from presentation to procedure start was 8.7 hours (IQR 5.7–15.9), with 50.0% of cases begun within 6 hours, and 100% of cases begun within 24 hours. Finally, for patients with NSTI, the median time from presentation to procedure start was 10.1 hours (IQR 5.6–17.5), with 55.6% of cases meeting the guideline for debridement within 12 hours.

Notably, the time elapsed between arrival at Hospital Y and procedure start varied between these acute conditions. For STEMI cases, the median time from arrival to procedure start was 0.2 hours (IQR 0.1–1.1). However, for patients with ALI and NSTI, patients waited a median of approximately 3 hours between arrival and procedure start (Table 4).

DISCUSSION

This study describes our institution’s early experience with a new hub-and-spoke health system model in which all patients requiring non-bedside procedures at the spoke hospital are transferred over a short geographic distance to the hub hospital for treatment. To our knowledge, this study is the first to explore the implications of this type of health system structure on access to emergency medical and surgical procedures. We found that, although more than half of patients were transferred within their assigned goal time, a substantial proportion of patients experienced transfer delays. For cases deemed highest priority, only 57.6% arrived within the goal of ≤ 2 hours, and for the second highest priority cases, 73.0% arrived within the goal of ≤ 8 hours. Furthermore, we found that the proportion of on-time transfers varied month to month, with periods of capacity strain at the hub hospital associated with lower rates of timely transfer from the spoke. Finally, in our subgroup analyses for three common emergency conditions (STEMI, ALI, and NSTI), we found that a substantial proportion of cases were not treated within the time frame recommended by professional guidelines.

Delayed access to emergency procedures can have grave repercussions for patient outcomes. Across a variety of conditions, studies have demonstrated that earlier intervention is associated with less morbidity and mortality. For example, a recent study examining outcomes after surgical repair of Type A aortic dissections found that mortality more than doubled for patients treated 8 to 12 hours after diagnosis compared to those treated within 8 hours.¹⁵ Similarly, for patients presenting with incarcerated abdominal wall hernias, the rates of complications and 30-day mortality were significantly increased when surgical intervention was performed more than 1 day after presentation.¹⁶ Moreover, for conditions such as stroke or ALI, patients may become ineligible for reperfusion procedures that would significantly improve quality of life if they are unable to access a capable facility within several hours of diagnosis.^13,17

However, although access to timely intervention is a key factor in optimizing patient outcomes, expertise is also essential. Recent work across multiple disciplines has shown a clear relationship between higher proceduralist case volume and improved outcomes.^7,9,18,19 In addition to the importance of expertise at the individual provider level, numerous studies have demonstrated that higher hospital case volume also matters, suggesting that institutional resources, clearly defined perioperative pathways, and interprofessional experience may further benefit patients.^20–24

Thus, for patients with emergency procedural needs, access to timely intervention and to a high-quality, high-volume center may represent competing priorities. The hub-and-spoke arrangement presents an appealing path toward achieving a balance of these priorities, as it has the potential to allow health systems to reduce low-volume specialty care at the spoke and continue to provide satisfactory patient outcomes if optimal transfer processes are in place. Yet as our current study demonstrates, even exceptionally short-distance transfers can lead to significant delays in access to time-sensitive interventions for emergency conditions.

In the limited literature that has examined the impact of transfer-associated delays for patients requiring time-sensitive procedures, the results for patient outcomes have been mixed. In one study of patients with stroke in New York, there was no difference in mortality between patients who initially presented to a hub and those who presented to a spoke.²⁵ In contrast, two national studies examining patients with ALI and patients requiring common general surgical operations found that those who were transferred to another facility for intervention had an increased like-lihood of mortality compared to those who were not.^26,27 One potential explanation for these disparate findings is the variability in time from diagnosis to intervention. In two studies examining transfer timeliness for patients requiring intervention for stroke and hip fracture, time from arrival at the spoke hospital to hub transfer varied widely.^28,29 In the present study, we similarly found substantial variation in the time it took to identify an available bed for patients requiring transfer as well as the time it took for patients to be picked up and transported. Thus, although hub-and-spoke restructuring can expand access to high-quality care, it requires clear operational protocols and oversight to ensure efficient treatment of time-sensitive conditions. Operational lessons may be learned from freestanding EDs that face similar challenges with regard to transfer of time-sensitive conditions requiring a higher level of care.^30,31

One particularly interesting finding from the current study is that the time from transfer request placement to arrival at the hub hospital was significantly shorter for patients with STEMI and NSTI compared to other patients with the highest priority transfer designation. This suggests that our institution’s transfer coordinators and transport teams are appropriately expediting these cases, but the mechanism of prioritization is not well-defined. It is possible that these conditions are better-recognized as true emergencies to the team members involved in the transfer process, given these conditions’ publicity in the lay media, compared to other conditions such as ALI. In addition, the markedly shorter wait time between arrival at Hospital Y and procedure start for STEMI patients compared to ALI and NSTI patients likely reflects the differences in arrival process, as STEMI can be taken directly to the catheterization laboratory rather than to an inpatient floor first.

To more fully understand the underlying reasons for these findings and to reduce the observed delays in time from diagnosis to procedure, our institution has convened an interdisciplinary quality improvement workgroup focused exclusively on procedural transfers from Hospital X to Hospital Y. The workgroup has begun to explore opportunities to reduce delays by standardizing transfer workflows and improving communication. In addition, the workgroup is developing an arrival process for surgical patients to mirror the STEMI patient process, such that surgical patients can be transferred directly to the perioperative area rather than to an inpatient floor first.

Limitations

This study has several limitations. First, as a single-institution study, generalizability is inherently limited. However, given the dearth of evidence regarding the implications of this type of hub-and-spoke arrangement for patients requiring emergency procedures, our experience serves as an example for other health systems considering a similar restructuring and illuminates the importance of planning and tracking the emergency transfer process. Second, although this study provides key metrics on transfer and procedure timing, it does not directly address the downstream consequences for morbidity and mortality in our cohort. Unfortunately, our dataset did not include information on patients’ preexisting comorbidities or granular information about disease severity at presentation, such as vital signs or laboratory data, which would be necessary to isolate the relationship between procedure delays and clinical outcomes.

CONCLUSION

As health systems consider restructuring the relationship between hospitals within close geographic proximity to conserve resources, the needs of patients presenting with emergency conditions should be carefully considered. Maintaining access to timely intervention has been a challenge for our institution under this novel hub-and-spoke arrangement, but the benefits of access to high-volume pro-ceduralists and experienced care teams has obvious value. In conclusion, we assert that the timeliness of emergency procedures may be an overlooked yet critical quality improvement priority for hub-and-spoke health systems.