Geographic Diffusion and Implementation of Acute Care Surgery: An Uneven Solution to the National Emergency General Surgery Crisis

Key Points

Question  In which communities is uptake of acute care surgery lagging, what are the characteristics of such communities, and why does it matter?

Findings  In this study pairing a national survey of hospitals with geographic information systems and census data, we found that acute care surgery implementation has occurred unevenly across the United States. Rural regions and communities with greater percentages of uneducated patients have had slower acute care surgery uptake.

Meaning  Acute care surgery, while a potential solution to our national crisis in access to care for general surgery emergencies, has affected vulnerable populations differentially, leaving rural and less educated areas at higher risk of lacking access to emergency general surgery care.

Importance  Owing to lack of adequate emergency care infrastructure and decline in general surgery workforce, the United States faces a crisis in access to emergency general surgery (EGS) care. Acute care surgery (ACS), an organized system of trauma, general surgery, and critical care, is a proposed solution; however, ACS diffusion remains poorly understood.

Objective  To investigate geographic diffusion of ACS models of care and characterize the communities in which ACS implementation is lagging.

Design, Setting, and Participants  A national survey on EGS practices was developed, tested, and administered at all 2811 US acute care hospitals providing EGS to adults between August 2015 and October 2015. Surgeons responsible for EGS coverage at these hospitals were approached. If these surgeons failed to respond to the initial survey implementation, secondary surgeons or chief medical officers at hospitals with only 1 general surgeon were approached.

Interventions  Survey responses on ACS implementation were linked with geocoded hospital data and national census data to determine geographic diffusion of and access to ACS.

Main Outcomes and Measures  We measured the distribution of hospitals with ACS models of care vs those without over time (diffusion) and by US counties characterized by sociodemographic characteristics of county residents (access).

Results  Survey response rate was 60% (n = 1690); 272 responding hospitals had implemented ACS by 2015, steadily increasing from 34 in 2001 to 125 in 2010. Acute care surgery implementation has not been uniform. Rural regions have limited ACS access, with hospitals in counties with greater than the 75th percentile population having 5.4 times higher odds (95% CI, 1.66-7.35) of implementing ACS than hospitals in counties with less than 25th percentile population. Communities with greater percentages of adults without a college degree also have limited ACS access (OR, 3.43; 95% CI, 1.81-6.48). However, incorporating EGS into ACS models may be a potential equalizer for poor, black, and Hispanic communities.

Conclusions and Relevance  Understanding and addressing gaps in ACS implementation across communities will be crucial to ensuring health equity for US residents experiencing general surgery emergencies.

All American individuals are susceptible to general surgery emergencies (eg, appendicitis and bowel obstruction); however, there exists inequity in access to emergency general surgery (EGS) care across the United States.^1-5 In 2006, the Institute of Medicine identified a national crisis in emergency care.⁶ Insufficient access to EGS care was cited as a major contributor owing to rising numbers of emergency department (ED) visits, ED closures, and a dwindling general surgery workforce.^6-15

Four million patients who visit EDs annually require EGS evaluation, with 2.3 million requiring admission.¹⁶ Once at the hospital, the fate of an EGS patient depends on the facility’s ability to provide nonelective general surgery, which depends on the availability of an on-call surgeon, an operating room (OR) that can function off-hours, and perioperative services. Because not all hospitals have this capacity, patients who present to their closest ED may not receive appropriate care.⁹

Historically, general surgeons have taken EGS call for both trauma and general surgery emergencies. However, in a national survey of ED directors, 37% reported inadequate EGS coverage.¹⁷ Citing largely lifestyle reasons, few medical students display interest in entering general surgery,¹⁸ and nearly 80% of general surgery residents specialize.^7,13 The shortage of general surgeons especially affects rural communities, where general surgeons tend to be older.¹⁰ As they retire, the difficulty of recruiting new surgeons risks hospital closures.¹⁴ Additionally, ED closings disproportionately affect vulnerable communities including those that are predominantly rural, poor, black, and Hispanic.^19,20

In response to this crisis, the structure of delivering EGS care is evolving from general surgeons taking call to an organized delivery system, acute care surgery (ACS), which has been proposed to encompass surgical critical care, trauma, and EGS.²¹ Despite having shown improvement in mortality, postoperative complications, time to the operating room, and length of stay, patterns of ACS implementation are poorly understood.^22,23 Moreover, the ideal components of an ACS service remain unclear.²³ There are no standardized implementation guidelines, triage algorithms, or ACS certification systems, nor is there a central database of hospitals with ACS.

We investigated the diffusion of ACS nationally and the sociodemographic characteristics of the communities where ACS has and has not been implemented. Quiz Ref IDWe hypothesized that vulnerable communities that might stand to benefit the most from ACS diffusion would have the slowest uptake.

A survey (eAppendix in the Supplement) to measure EGS processes was developed, based on qualitative research²⁴ and a pilot survey.²² Because there is no accepted definition of ACS, we used our preliminary research to generate a questionnaire item to capture whether ACS, based on local standards, had been implemented.²⁴ In this study, we specifically address responses to the item, “my hospital’s overall approach to EGS is… (1) A dedicated clinical team whose scope encompasses EGS (±trauma, ±elective general surgery, and ± burns), (2) a traditional approach with an ad hoc general surgeon on call schedule, or (3) other.” Those who responded with the first option (ACS hospitals) were also queried about the year ACS was implemented (and month, if implemented in 2015).

To identify acute care general hospitals providing EGS care to adults (≥18 years), we used the American Hospital Association (AHA) Annual Survey of Hospitals (2013)²⁵ database to search for “acute care hospital,” “surgical services,” “ED,” and at least 1 “OR.” Exclusion criteria included hospitals without an ED or an OR; specialty hospitals (eg, children’s, orthopedic, and cardiac), hospitals in US territories; and hospitals inaccessible to the general public. We identified 3322 hospitals reported by the AHA to have both an ED room and an OR. Provision of constant access to an ED, an OR, and a general surgeon at these hospitals was confirmed through websites and direct contact, if necessary. Four hundred sixty-nine hospitals lacking general surgery coverage, an ED, or an OR were excluded. After survey implementation to the remaining 2853 hospitals, 42 hospitals replying that EGS was not provided were also excluded. The final sample consisted of 2811 EGS hospitals.

Targeting in a greater than 50% hospital response rate, we identified 2 senior surgeons (primary and secondary) at each hospital, with plans to use secondary contacts if the primary surgeon failed to respond. We targeted senior surgeons because our prior research suggested they would have the greatest insight into the organizational details of EGS coverage at the hospital. If only 1 surgeon was identified, we used the chief medical officer as the secondary contact. Contacts were identified using online and telephone inquiries, surgical society websites, and AHA data. If the initial survey implementation yielded new contact information, the primary surgeon was recontacted during the second survey implementation.

Between August 2015 and October 2015, the survey was sent to primary surgeons at all hospitals in our sample using a hybrid paper/electronic approach. The paper survey, along with a small incentive, was mailed, and an email reminder was sent 3 days, 2 weeks, and 6 weeks later. Those without a valid email (n = 1354; 39.5%) received a second paper survey without incentive 2 weeks later. With only 38.7% of hospitals represented by October 2015, the same survey implementation was repeated between October 2015 and December 2015 targeting secondary surgeons (n = 1497), chief medical officers (n = 222), or primary surgeons with new contact information (n = 207) for the remaining 1926 hospitals.

This study was deemed exempt by the University of Massachusetts institutional review board because the survey content was not related to sensitive information and response to the survey indicated consent to have responses analyzed in aggregate form only.

Characteristics (from 2015 AHA data) of nonrespondent and respondent hospitals and of ACS and non-ACS hospitals among respondents were compared using t tests, nonparametric tests, and χ² tests using SAS, version 9.4 (SAS Institute). Thirteen hospitals represented in our survey were not found in the 2015 AHA database (3 closed, 5 merged/renamed, and 5 unknown); we used the 2013 AHA data for these 13 hospitals that were operational at the time of our survey.

We analyzed trends in ACS uptake over time. Respondent hospitals were overlaid on US maps using Maptitude Geographic Information Systems software (Caliper Corp). To study population-level characteristics, 2010 US Census data on race, ethnicity, poverty (<200% below federal poverty level), education level (≥25 years with minimum of bachelor’s degree), and insurance coverage (≥18 years without health insurance) were overlaid on maps of hospital distribution. To create percentage groupings for black race, Hispanic ethnicity, education level, insurance status, and poverty, the Fisher-Jenks Optimal Breaks algorithm was used. This method groups data into classes such that within-group variance is minimized, while difference in mean of any single group compared with the mean of other groups is maximized.²⁶

Acute care surgery uptake by 2015 was assessed at the county level based on population from 2010 Census data and the previously mentioned sociodemographic characteristics by calculating the 25th and the 75th percentile among all US counties for each sociodemographic variable. We determined odds of having implemented ACS in a 75th, 51th to 74th, or 26th to 50th percentile county vs a 25th percentile county for a given variable. The P value level of significance was set at .05 for a 2-sided test of significance.

Of the 2811 EGS hospitals surveyed, 1690 responded (60% response rate). Respondent and nonrespondent hospitals did not differ significantly in rurality. Quiz Ref IDHowever, hospitals represented in the survey results were more likely to be affiliated with medical schools, teaching hospitals, and nongovernmental ownership (eTable 1 in the Supplement).

Quiz Ref IDOf the responders, 94.4% answered the question on ACS implementation as of 2015, with 16.1% responding yes, 72.5% reporting an ad hoc general surgeon on-call schedule, and 5.8% responding “other” (ie, patients are transferred if a general surgeon was not available, 1 general surgeon on call, or hybrid model). Of 244 hospitals reporting implementation year, 32 were established before 2001, with the earliest being 1980 and the most recent being September 2015. Figure 1 shows the change in ACS implementation over time nationally (2.1% ACS in 2001, 2.4% ACS in 2002, 2.8% ACS in 2003, 3.1% ACS in 2004, 3.7% ACS in 2005, 4.2% ACS in 2006, 4.8% ACS in 2007, 5.7% ACS in 2008, 6.4% ACS in 2009, 7.8% ACS in 2010, 9.2% ACS in 2011, 10.8% ACS in 2012, 12.5% ACS in 2013, 14.2% ACS in 2014, and 15.3% ACS in 2015; P value for trend <.001).

Geographically, the fewest ACS hospitals were in the East South Central region, while the most were in South Atlantic region; however, these geographic differences were not statistically significant. Quiz Ref IDAcute care surgery hospitals were more likely to be urban, greater than 500 beds, noncritical access, and teaching hospitals, with all P values <.001 (eTable 2 in the Supplement).

Acute care surgery hospitals (n = 272) and non-ACS hospitals (n = 1324) were mapped over the population density (Figure 2). Areas with a low population density were unlikely to have ACS implementation, with only 1.0% of counties with less than 50 000 persons (n = 2242) having at least 1 ACS hospital. In contrast, of counties with greater than 500 000 persons (n = 133), 63.2% had at least 1 ACS hospital. Hospitals in counties with greater than the 75th percentile population (>65 854 individuals) had 5.4 times higher odds (95% CI, 1.66-7.35) of implementing ACS than hospitals in counties with less than 25th percentile population (<11 096 individuals) (Table).

Figure 3 shows ACS hospitals (n = 272) by county-level percentage of African American and Hispanic population. Quiz Ref IDHospitals in counties greater than the 75th percentile of African American population (10.2%) had 4.6 times higher odds of implementing ACS compared with hospitals in counties less than 25th percentile (0.6%) (Table). Similar results are apparent for the distribution of Hispanic populations and hospitals that implemented ACS.

Figure 4 shows ACS hospitals by county-level rates of uninsurance, persons living in poverty, and adults with a college degree. While insurance status was not associated with ACS implementation, communities with greater proportion of residents with a college degree or who are below poverty level are more likely to have ACS hospitals (Table).

In this study, we found that while ACS implementation is increasing, uptake is uneven across regions. Yet early reports from hospitals adopting ACS suggest a number of improvements in quality, costs, and outcomes including optimal OR use, reduced length of stay, and improved mortality.^27-36 Furthermore, regional studies indicate that patients with general surgery emergencies may have specific demographic, comorbid, and physiologic risk factors that warrant early referral to ACS hospitals with constant OR preparedness and specialized critical care resources.^37-44 That substantial disparities in uptake exist by population density and education level suggests widening gaps in access to high-quality and timely EGS care in rural counties and counties with less educated populations.

A 2016 study⁴⁵ of EGS patients transferred to a rural tertiary referral center found that patients traveled as far as 315 miles from 29 different hospitals.⁴⁵ Despite extensive travel, 31% of transferred patients did not undergo intervention.⁴⁵ Moreover, those who did undergo surgery did so for relatively simple and common procedures. Compared with urban patients, rural patients have also been found to have greater odds of mortality following EGS⁴; excess travel time may be a factor in this outcomes disparity.^43,44,46 As ACS implementation increases, its ability to diffuse to less populated areas and improve outcomes for rural residents with general surgery emergencies will have to be specifically considered.

While one impetus for proposing ACS was a crisis in the general surgery workforce,⁷ current training paradigms may not be meeting the needs of rural residents who are lacking surgeons or hospitals for EGS care. Reports indicate that there are 6.5 surgeons to 100 000 individuals in urban areas compared with 4.7 to 100 000 in rural areas.¹⁰ As general surgeons retire without surgeons to replace them, lack of access to EGS care will be exacerbated.^10,14 Moreover, ED closures have disproportionately affected rural areas,¹⁹ and without an ED, the hospital cannot provide EGS. Meanwhile, a survey of graduated ACS fellows found that only 7% practice at rural hospitals.⁴⁷ That those who are trained in ACS are not practicing in areas with the greatest disparities in access to EGS care likely explains our findings that ACS uptake in rural areas is lacking.

We have previously shown that EGS deserts, ie, areas with limited or no access to a hospital providing EGS care, exist across the nation and are more likely to be home to vulnerable communities.⁴⁸ While some of these patterns of disparities are mirrored in our results regarding ACS diffusion, uptake appears higher at hospitals with existing EGS capabilities located in more densely populated areas that are both home to minority populations and those living in poverty while also attracting more highly educated residents. Owing to this disproportionate diffusion of ACS, some of the population is benefiting by receiving EGS care with ACS models, some of the population is receiving EGS care with non-ACS models, and the some of the population is experiencing poor outcomes without timely access to high-quality EGS care. Our findings of uneven uptake of ACS demonstrate the need for establishing criteria, as has been done for injured patients needing trauma care, for triage, and for regionalization of care for patients living in more remote areas who experience general surgery emergencies.

Predominantly black and Hispanic communities are often EGS deserts.⁴⁸ Somewhat paradoxically, we have shown when such communities do have access to EGS, they are more likely to have access to an ACS hospital. This suggests that minority individuals living in densely populated areas, perhaps in proximity to large, tertiary care teaching hospitals, benefit from disproportionate implementation of ACS at these centers. This might explain why EGS outcomes research regarding Hispanic individuals has been mixed despite higher relative risk of worsening access to EDs.¹⁹ Nevertheless, Hispanic individuals are found to have among the lowest complication and mortality rates following most EGS procedures.⁴⁹ However, Hispanic individuals are less likely to receive minimally invasive surgery and be treated by a top-decile surgeon or at a high-volume hospital.⁵⁰ Conversely, African American EGS patients consistently have poorer outcomes across all domains and in all settings.^2-4 The reasons for this are unclear. Further research is necessary to fully understand how access and outcomes of nonwhite patients are affected by ACS implementation and whether minority individuals residing in rural areas are even more vulnerable than their white rural counterparts.

We found that county-level insurance coverage was not associated with ACS implementation. In contrast to our predictions, respondent hospitals in counties with greater levels of poverty had higher odds of implementing ACS compared with their counterparts. This may also be the result of dense pockets of urban poor living in proximity to large tertiary care centers. Among EGS patients nationally, insurance status has been found to be independently associated with severity of disease at presentation,⁵¹ complications,¹ and mortality.¹ Moreover, low socioeconomic status is associated with higher mortality and complications for EGS.¹ It has also been shown that following EGS, uninsured patients have 25% higher odds of dying, while the wealthiest patients have 14% lower odds of dying.⁴ While our previous research showed that communities with high rates of poverty and uninsured individuals are more likely to lack access to a hospital providing EGS care,⁴⁸ these findings differed. Thus, as ACS implementation continues, it will be important to consider opportunities for outreach to poor and uninsured patients in areas where there are no other EGS-capable hospitals.

Our findings must be considered with an understanding of the limitations of our research. Insurance data used were from the 2010 US Census, before the Affordable Care Act was passed, and may not be an accurate reflection of the United States in 2017. Also, by analyzing demographics of the communities in which the hospitals are located, we assumed patients receive emergency care locally. Finally, given some differences between responding and nonresponding hospitals, our findings regarding the national uptake of ACS may not be generalizable; however, because we found that tertiary, teaching, larger-sized hospitals were more likely to adopt ACS, our final assessment of where ACS is diffusing will be minimally affected.

Acute care surgery was proposed using lessons from organized trauma teams to address simultaneous decline in EGS care access and waning interest in trauma/critical care careers.²¹ While an explicit intent to address the EGS needs of vulnerable communities was not evident in the original studies proposing ACS, this study, with its national view on the diffusion and differential availability of ACS, provides the societal context for a growing problem in access to EGS care that has been apparent to surgeons for some time now. While, nationally, trauma patients have benefited from standardized regionalization of care, EGS patients have been left behind.^52,53 Our findings indicate that persons with a general surgery emergency in densely populated urban areas regardless of race, ethnicity, and socioeconomic status will be most likely to have access to an ACS hospital and presumably the many process and outcome benefits associated with this model of care. These findings have implications on how to increase access for EGS patients, who, similar to trauma patients, cannot optimize comorbidities and marshal socioeconomic resources prior to seeking care.

Our past qualitative research suggests there is not a “one-size-fits-all” model of ACS.⁵⁴ Thus, as the diffusion of ACS continues, it will be critically important to adjust these sizes to augment access to EGS care among residents of rural and less educated communities, while bearing in mind, as with numerous other complex diseases, the beneficial association between volume and outcomes for EGS.⁵⁵ Delivering EGS care through tiered, regionalized programs, as has been done for trauma centers and neonatal intensive care units, may be a solution. While not all hospitals will be in a position from a resource or a volume perspective to provide the highest level of care, agreed-on guidelines for basic and intermediate capabilities, as with leveling of trauma centers and neonatal intensive care units, will be beneficial to both health systems and patients in the regions they serve. The current alternative is to rely on a decades-old paradigm of an all-encompassing general surgeon on call who may no longer exist in many areas. Such a tiered system would necessarily be optimized with triage criteria to evaluate patients based on both the presenting disease, severity of illness, and comorbidities as well as by preexisting agreements for expedited transfer to higher levels of care when existing resources cannot meet patient needs. There may also be a role for telemedicine, as has been implemented for stroke care,⁵⁶ to identify which patients truly need transfer to a higher level of care.

Importantly, where communities and policy makers are actively engaged in ensuring timely access and highest quality of care for injured patients, newborns, and stroke patients, no such extrainstitutional interests appear to have taken hold for EGS patients, to our knowledge. As we move forward, it is vital to ensure that ACS implementation across hospitals, a process that has heretofore been led by well-intentioned surgeons, includes stakeholders and policymakers in communities that are at particularly high risk of lacking access to high-quality EGS care.

Corresponding Author: Heena P. Santry, MD, MS, Department of Surgery, University of Massachusetts Medical School, 55 Lake Ave North, Worcester, MA 01655 (heena.santry@umassmemorial.org).

Accepted for Publication: June 25, 2017.

Published Online: October 4, 2017. doi:10.1001/jamasurg.2017.3799

Author Contributions: Dr Santry and Ms Ayturk had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Khubchandani, Santry.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Khubchandani, Ayturk, Santry.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Ingraham, Daniel, Ayturk.

Obtained funding: Santry.

Administrative, technical, or material support: Ingraham, Santry.

Supervision: Kiefe, Santry.

Conflict of Interest Disclosures: None reported.

Funding/Support: This research is supported by grant R01HS022694 from the Agency for Healthcare Research Quality (Dr Santry), grant R01MH112138 from the National Institutes of Health/National Institute of Mental Health (Dr Kiefe), grant U54 RR 026088 from the National Institutes of Health/National Center for Advancing Translational Sciences (Dr Kiefe), and grant ME-1310-07682 Patient-Centered Outcomes Research Institute (Dr Kiefe).

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.