Training in Trauma Surgery

Abstract

Objective:

To describe outcomes from a clinical trauma surgical education program that places the board-eligible/board-certified fellow in the role of the attending surgeon (fellow-in-exception [FIE]) during the latter half of a 2-year trauma/surgical critical care fellowship.

Summary Background Data:

National discussions have begun to explore the question of optimal methods for postresidency training in surgery. Few objective studies are available to evaluate current training models.

Methods:

We analyzed provider-specific data from both our trauma registry and performance improvement (PI) databases. In addition, we performed TRISS analysis when all data were available. Registry and PI data were analyzed as 2 groups (faculty trauma surgeons and FIEs) to determine experience, safety, and trends in errors. We also surveyed graduate fellows using a questionnaire that evaluated perceptions of training and experience on a 6-point Likert scale.

Results:

During a 4-year period 7,769 trauma patients were evaluated, of which 46.3% met criteria to be submitted to the PA Trauma Outcome Study (PTOS, ie, more severe injury). The faculty group saw 5,885 patients (2,720 PTOS); the FIE group saw 1,884 patients (879 PTOS). The groups were similar in respect to mechanism of injury (74% blunt; 26% penetrating both groups) and injury severity (mean ISS faculty 10.0; FIEs 9.5). When indexed to patient contacts, FIEs did more operations than the faculty group (28.4% versus 25.6%; P < 0.05). Death rates were similar between groups (faculty 10.5%; FIEs 10.0%). Analysis of deaths using PI and TRISS data failed to demonstrate differences between the groups. Analysis of provider-specific errors demonstrated a slightly higher rate for FIEs when compared with faculty when indexed to PTOS cases (4.1% versus 2.1%; P < 0.01). For both groups, errors in management were more common than errors in technique. Twenty-one (91%) of twenty-three surveys were returned. Fellows’ feelings of preparedness to manage complex trauma patients improved during the fellowship (mean 3.2 prior to fellowship versus 4.5 after first year versus 5.8 after FIE year; P < 0.05 by ANOVA). Eighty percent rated the FIE educational experience “great -5” or “exceptional– 6.” Eighty-five percent consider the current structure of the fellowship (with FIE year) as ideal. Ninety percent would repeat the fellowship.

Conclusion:

The educational experience and training improvement offered by the inclusion of a FIE period during a trauma fellowship is exceptional. Patient outcomes are unchanged. The potential for an increased error rate is present during this period of clinical autonomy and must be addressed when designing the methods of supervision of care to assure concurrent senior staff review.

Methods of training fellows in trauma care vary widely. Little objective clinical data is available that validate current training models. We describe our experience with a model of clinical trauma education that places the fellow in the role of the attending during the latter half of their training. Through the analysis of data from an innovative performance improvement process, outcome measures, and a survey of trainees, we have evaluated safety and educational experience.

Trauma fellowships were begun in the United States in the 1970s to give additional training for general surgeons with an interest in traumatology. Trauma fellowship programs have steadily increased in number since that time. As the number of programs has increased, so has their variability. Although there have been some published recommendations for the structure of these programs,^1,2 there is still a wide range in program structure and content. The Accreditation Council for Graduate Medical Education (ACGME) offers accreditation for surgical critical care (but not trauma) fellowships. Currently, there is no accreditation board for trauma surgery, although the American Board of Surgery has discussed the possibility of creating a special certificate for trauma surgery. This has not materialized. Some data indicate that trauma programs are more attractive when combined with accredited surgical critical care fellowships.^3-6 Those programs that offer accredited surgical critical care training also vary considerably. Most have some combination of clinical patient care, research, resident teaching, and administrative education.⁷ Some fellowships last only 1 year, focusing on critical care with a small addition of trauma care. Others are 2 years long; most frequently 1 year dedicated to SCC followed by 1 dedicated to trauma. Program directors have constructed their respective programs to highlight the strengths of their individual program on the basis of their specific resources. Little objective clinical data are available that validates any of the current training models.

The University of Pennsylvania School of Medicine has offered a combined fellowship in traumatology and surgical critical care since 1989. It is a 2-year program, with portions of each year devoted exclusively to either trauma or critical care. Currently, the first year consists of 4 months of adult critical care, 1 month of pediatric critical care, 1 month of pediatric trauma, and 6 months of adult trauma in the role of “junior fellow.” The second year consists of 5 months of adult critical care and 7 months of adult trauma in the role of “senior fellow.” While on the trauma service during the second year, the senior fellow is the Fellow-in-Exception (FIE). These FIEs are credentialed and appointed as active members of the medical staff,⁸ with full operating room and admitting privileges. On the trauma service, they take on the role of the trauma attending.

We hypothesize that our model is safe for patient care while providing a superior clinical experience for the trainee. We describe our experience with this model of clinical trauma education.

MATERIALS AND METHODS

Fellowship Structure

The University of Pennsylvania School of Medicine fellowship in traumatology and surgical critical care has been ACGME-approved since 1994. Board-eligible or board-certified general surgeon trainees take part in an integrated 2-year program, with portions of each year devoted exclusively to either trauma or critical care. Currently, the first year consists of 4 months of adult critical care, 1 month of pediatric critical care, 1 month of pediatric trauma, and 6 months of adult trauma. The second year consists of 5 months of adult critical care and 7 months of adult trauma as an FIE.

These FIEs take in-hospital call and oversee the inpatient service as a trauma attending in a fully integrated faculty coverage plan (trauma faculty take in-hospital call and cover the in-patient service in an identical fashion.) Out-of-hospital backup is always available for the in-hospital surgeon, regardless of whether the in-house surgeon is FIE or faculty. All patient contacts, admissions, operations, or major decisions for a 24-hour period are reviewed at a morning report, which is held 365 days a year. In this forum, senior faculty and fellows not involved with the case or its management, critically review digital radiographic images and reports, laboratory studies, operative procedures, and decision-making, essentially evaluating all care in a near-concurrent fashion.

Quantitative Data: Registry and Performance Improvement Data

For this study, data were abstracted from 2 computerized databases for a 4-year period from July 1998 to June 2002. The trauma registry (Collector Trauma Registry, Digital Innovations, Inc., Forest Hills, MD) was used to abstract total patient contacts, Pennsylvania Trauma Outcome Study (PTOS) patients (Table 1), mechanism of injury, injury severity score (ISS), and operations in a provider-specific fashion. Of note, provider-specific data for operations (both acute and nonacute) were available for the last 3 years of the study only. TRISS analysis was performed on patients when all necessary data were available. Unexpected deaths and unexpected survivors were identified using accepted definitions.⁹

TABLE 1. Pennsylvania Trauma Outcome Study Patients

Second, performance improvement (PI) data were abstracted in a similar provider-specific fashion from a securely maintained computerized database (Table 2). The PI database tracks provider-specific data on all deaths and occurrences (eg, audit filters, complications, clinical management guideline [CMG] violations). Occurrences are a combination of Pennsylvania Trauma System Foundation (PTSF) defined incidents as well as institution-specific trauma PI filters. Deaths are peer reviewed by all providers, assigned to a specific provider when appropriate, and classified as nonpreventable, potentially preventable, or preventable.¹⁰ Occurrences and audit filters are also evaluated in the PI process. When possible errors are identified, the care is peer reviewed by all providers and assigned to a specific provider when appropriate. Errors were divided into 2 categories: errors in technique and errors in management.

TABLE 2. Glossary of Performance Improvement Terms¹⁰

All quantitative data was analyzed as 2 groups: faculty trauma surgeons and FIEs. The alternative critical ratio test, Fisher exact test, and t test provided between-group comparisons for questions of interest. In addition, a power analysis was performed to better analyze the significance of survival comparisons between groups.

Qualitative Data: Past Fellow Survey

A 25-item survey was created specifically for this study. The survey was completed anonymously. Representative survey questions are presented in Table 3. Questions focused on perceptions of clinical management skill and fellowship logistics, as well as strengths and weaknesses of the program. Where appropriate, a 6-point Likert scale was employed to evaluate perceptions of training and experience. All fellows who had completed the 2-year fellowship in traumatology and surgical critical care at the University of Pennsylvania were electronically forwarded the survey employing an on-line tool (Zoomerang.com, MarketTools Inc, Mill Valley, CA). Questions of interest were statistically analyzed using a Wilcoxon signed rank test.

TABLE 3. Survey Tool

Institutional Approval

Institutional Review Board approval was obtained from the University of Pennsylvania.

RESULTS

Quantitative Data: Registry and Performance Improvement Data

Eleven fellows completed the 2-year fellowship during the study period and therefore comprise the FIE group. During this time, a total of 10 trauma faculty members (mean length of service as trauma faculty at study midpoint was 6.7 years) participated in trauma service coverage and provided the majority of clinical care to injured patients at the Hospital of the University of Pennsylvania. These surgeons comprise the trauma faculty group.

Trauma registry data are presented in Table 4. Both groups saw predominantly patients injured from a blunt mechanism. Approximately one quarter of all patients sustained penetrating trauma; more than 75% of these had sustained gunshot wounds. Objective measures of injury severity (PTOS patients, mean ISS, ISS > 16) were similar between groups. Of note, the 11 FIEs evaluated an average of 171 acutely injured patients (contacts) during the second year of their fellowship. Data on all operations performed on trauma patients during the latter 3 years of the study are displayed in Table 5. FIEs performed an average of 52.5 acute and nonacute operations. Indexed to the number of trauma patient contacts evaluated, fellows-in-exception performed significantly more operative procedures than faculty.

TABLE 4. Trauma Registry Data

TABLE 5. Trauma Operations (7/99–6/02)

Outcome data from the 2 groups are presented in Table 6. The overall mortality rate (deaths/PTOS patients) was 10.4% for the time period studied. No difference between study groups was identified. The study had a power of 80% to detect a 4% difference in survival between the 2 groups with an α ≤ 0.05. Peer review judgments on all deaths were recorded. No preventable deaths were identified during the study period; 3.2% of the deaths were determined to be potentially preventable. Again, no differences were identified when comparing groups. TRISS analysis was performed when all necessary data were available. The incidence of TRISS unexpected deaths was higher in the FIE group, but this did not reach statistical significance. In addition, a similar analysis of TRISS unexpected survivors failed to demonstrate a significant difference.

TABLE 6. Outcome Data

Complications identified in the PI process were also critically evaluated. Peer review judgment identified provider-specific errors in 93 instances (see Table 7). Provider-specific errors were infrequent but statistically more common in the FIE group. These errors tended to more frequently fall into the category of error in management than error in technique for both groups. The only statistically significant difference in specific errors was a higher incidence of delays to the operating room for the FIE group (5 of 27 versus 1 of 37 for the faculty group; P < 0.05).

TABLE 7. Provider-Specific Errors

Qualitative Data: Past Fellow Survey

Of the 23 surveys electronically sent, 21 surveys (91%) were returned. Fellows had completed their trauma training an average of 5.2 years prior to answering the survey. Thirty percent of respondents had some additional surgical experience (ie, other fellowship, surgery practice) before their trauma fellowship. Using a 6-point Likert scale, fellows’ feelings of preparedness to manage complex trauma patients improved during the fellowship (Fig. 1; mean, 3.2 prior to fellowship versus 4.5 after first year versus 5.8 after FIE year; P < 0.05). The FIE educational experience was rated “great -5” or “exceptional -6” by 80% of respondents. Eight-five percent of respondents consider the current structure of the fellowship (with the FIE period) as ideal. The other 15% suggest shortening the fellowship by 6 months, dropping the “junior trauma fellow” period. No past fellows suggested that the FIE period be omitted. Ninety percent of past fellows who responded would repeat the fellowship as it is currently structured.

FIGURE 1. Perceptions of preparedness to manage complex trauma patients. Response options ranged from “not at all –1” to “completely –6” (See Table 3). Mean scores improved in a statistically significant fashion over the course of the 2-year fellowship (P < 0.05 by Wilcoxon signed rank test).

DISCUSSION

The Maryland Institute for Emergency Medical Services offered the first formalized trauma fellowship in 1975. The 1-year fellowship focused on the care of the multisystem trauma patient, with an emphasis on intensive care unit management.⁶ Today, there are approximately 54 trauma fellowships in the United States.¹¹ The fellowships vary in length (1or 2 years or longer), as well as training emphasis (clinical or research). Many are combined with an ACGME-approved training program in Surgical Critical Care. Although “Guidelines for Trauma Care Fellowships” was published in 1992,² not all fellowships are guideline-compliant.¹² Little objective clinical data are available that validate current training models.

The University of Pennsylvania School of Medicine fellowship in traumatology and surgical critical care has been in place for more than 14 years. Board-eligible or board-certified general surgeon trainees take part in an integrated 2-year program, with portions of each year devoted exclusively to either trauma or critical care. While assigned to trauma for a total of 7 months during his/her second year, the senior fellow is the FIE, functioning as an attending on the trauma service. Although the FIE year has been very well received by our fellow trainees (during exit interviews as well as in trainee semi-annual evaluations of the fellowship), we had never critically evaluated the experience. This review represents the first performance-based evaluation of a trauma fellowship. To our knowledge, no other use of provider-specific trauma registry and PI data has been used in this way to evaluate surgical trainees or training programs.

The continued evolution of our PI process has allowed us to critically evaluate our fellowship program. Beginning in July 1998, provider-specific data was collected on all deaths and occurrences in a password-protected computerized database. The PI process is concurrent, with multiple sites of occurrence notification in both the inpatient and outpatient arenas. A full-time RN with more than 20 years experience is the trauma PI coordinator and has oversight authority on all matters related to quality assurance. The RN is supported by a PI medical director, who is a trauma surgeon, but not the trauma director or chief of service. The process is also peer reviewed. On a regular basis, all deaths and selected (prescreened) occurrences are discussed in an interdisciplinary forum. After care is reviewed, decisions are made on management. When appropriate, provider-related issues are identified. Specific providers are counseled and educated on an individual case basis or if a trend is identified. In addition, at the end of the year provider-specific summaries of trauma service activity gathered from trauma registry, and PI databases are reviewed by the chief of service and the trauma program director and distributed to individual providers. Subsequently, adjustments to the program are made according to any trends identified.

The current study provides objective data to evaluate the FIE portion of the fellowship (no data are provided for the first year of the trauma fellowship). FIEs each performed on average more than 170 acute patient evaluations and 50 operations on trauma patients over their 7-month experience. In addition, many more non-acute patient contacts occurred while the FIE oversaw the inpatient service. Although standards are in place for patient volumes during the surgical critical care portion of the fellowship (minimum of 5 intensive care unit [ICU] patients per day per fellow),¹³ no such standards exist for trauma fellowships.

Registry data comparing the experience of FIEs to that of the trauma faculty demonstrate few differences. Nearly 50% of patients in both groups met criteria for submission to the Pennsylvania Trauma Outcome Study (Table 1). These patients are considered by the PTSF to be serious or multiply injured patients. In addition, approximately 36% of these PTOS patients (in both groups) were also considered severely injured as defined by an ISS > 16. FIEs performed significantly more total operations on trauma patients over the last 3 years of the study, when provider-specific data for all operations were available. This finding was not surprising. Our FIEs are encouraged to be involved in many non-acute procedures on trauma patients throughout the normal workday. In addition, they are readily available to help during multiple patient resuscitations and operations whenever encountered. Still, the number of operations performed by the FIEs (as well as the trauma faculty) on trauma patients is low, underscoring the non-operative nature of modern trauma care. This is clearly a national phenomenon, discussion of which is beyond the scope of this manuscript.^14-18 In response to this issue, we have added a non-trauma emergency surgery component to the trauma service. This has markedly increased overall operative exposure for residents, junior fellows, FIEs, and faculty on the trauma service at our institution.¹⁹ Ninety percent of our prior fellows feel that the addition of urgent and emergency surgery would make a fellowship in trauma more attractive.

Analysis of registry and PI data has allowed us to examine patient safety as it relates to our fellowship structure. Overall mortality rates did not differ between groups. Of note, 70% of all deaths in our trauma center occur in the trauma resuscitation area or operating room, usually minutes after arrival. All deaths were peer reviewed in an interdisciplinary forum. The incidence of preventable and potentially preventable deaths was similar between groups. TRISS analysis of registry data also demonstrated no statistically significant difference in the incidence of TRISS unexpected deaths (or survivors) when comparing FIEs to trauma faculty. Other outcome measures, such as length of stay, were not examined. Although ICU and hospital length of stay data are tracked, the group practice model employed at this center makes it impractical to assign length of stay to a specific provider.

Occurrences were evaluated through our PI process. Most of occurrences were not specifically related to a single provider or were not classified as errors. Provider-specific errors were separated into 2 groups by provider status (faculty or FIE). Overall errors were more common in the FIE group. Delays to the operating room were the only error observed statistically more frequently in the FIE group as compared with the trauma faculty.

The identification of FIE (and faculty) errors in a near-concurrent fashion is of paramount importance when optimizing patient safety. This most regularly and frequently occurs at the morning report. At this forum, which is held 365 days a year, all patient contacts for a 24-hour period are reviewed. Senior faculty and other fellows critically review radiologic and laboratory studies, operative procedures, and decision-making on all new contacts as well as active existing patients. When necessary, immediate changes to the care plan are instituted, minimizing impact to patient outcome, morbidity, or cost. Ongoing care is also critically reviewed in a concurrent fashion by our PI nurse coordinator. CMG compliance is tracked and, when unexplained deviations are identified, corrective action is implemented. The exact number of times a change is made is not tracked, but we estimate that these changes are frequently minor adjustments and less commonly major changes in patient management. Immediately after morning report, bedside rounds are conducted daily. The faculty (or FIE) in charge of the inpatient service sees all patients with the housestaff on the trauma teams. Our perception has been that fewer changes in the care of new patients are made at the bedside in comparison to at the morning report forum.

The commitment to fellow (and resident) education is an important component of the ongoing PI process. In addition to individual case reviews, educational sessions focused on issues identified through the PI process work to minimize unwanted repetition of common errors. The near-concurrent peer review of trauma resuscitations at morning report, coupled with more formal focused educational sessions, may explain, in part, the improved outcomes seen at level I trauma centers with fellowship programs.²⁰ In our own program, besides being the cornerstone of education, morning report serves as the means to maximize patient safety on a busy trauma service.

From an educational standpoint, the results of our survey were largely positive. Fellows’ feelings of preparedness to manage complex trauma patients significantly improved during the fellowship. Eighty percent rated the FIE educational experience “great - 5” or “exceptional – 6,” and 85% consider the current structure of the fellowship (with FIE year) as ideal. Past fellows were also asked to rate strengths and weaknesses of the FIE experience. Fellows clearly found the added clinical experience (100% “agree –5” or “strongly agree –6”), and they graded transition from training to independence (95% agree or strongly agree) to be strengths of the program. Similar to a survey of 48 trainees in trauma fellowship programs throughout the country by Knuth et al⁴, our past fellows also felt that a relative lack of operative cases was an area of weakness for the FIE experience (31% agree or strongly agree). Lost income (37% agree or strongly agree) and additional time commitment (16% agree) were also mentioned. Still, 90% of past fellows would repeat the fellowship as it is currently structured. This is slightly greater than the 78% rate reported in Knuth’s survey.⁴

The current study has a number of weaknesses. It is a retrospective review of clinical care in a single institution. Outcome measures examined are limited. The peer review PI process employed is internal and has not been externally validated (except for intermittent site surveys by the PTSF) or translated to another institution. The grouping of providers into 2 distinct groups may also introduce flaws into our analysis. However, our year-end review of provider-specific summaries has to date failed to identify individual outliers. As a result, we feel our analysis of quantitative between-group data is justified. The survey tool was only sent to past fellows from the home program, and the potential for bias when participants answer questions about their own training program may be present. Still, we feel that this manuscript presents a unique insight into a trauma fellowship program. For our own faculty, as surgical educators and clinicians, we feel this research effort has served as an internal PI review of our own fellowship.

In summary, we have described a paradigm for trauma fellowship training. Senior fellows functioning in the role of trauma attending independently provide a portion of the care rendered to injured patients at a level I trauma center. Close faculty monitoring and provider-specific PI surveillance are employed to assure patient safety and optimize outcome. As a result, measured outcomes are equivalent to those seen with care rendered by faculty surgeons, and educational experience for the trainee appears to be maximized.

ACKNOWLEDGMENTS

The authors would like to thank Janet McMaster, RN (Trauma PI Coordinator), and Mary Kate Fitzpatrick RN, MSN (Trauma Clinical Administrator), for their ongoing dedication to the trauma program at the University of Pennsylvania. In addition, the authors would like to thank Seema S. Sonad, PhD, for her help with the design of the survey tool and Charles C. Branas, PhD, for his statistical review.

Discussion

Dr. L. D. Britt (Norfolk, Virginia): Thank you for allowing me to discuss this excellent paper. I want to thank the authors for sending me the manuscript several days in advance. I want to commend them on a very well-written manuscript that forms a possible template for performance improvement, evaluation, and outcome assessment.

With such stated variability in the existing trauma fellowships, there is a need to have some objective method to validate the current training models. The authors have taken on such a complex task and appropriately highlighted some of the inherent weaknesses of a retrospective review.

The objective of this study was to summarize data collected on a clinical trauma surgical program that places the Board-eligible, Board-certified fellow in the role of the attending surgeon during the latter half of the two-year program, and the authors note that in 10 instances, 3.5%, that a death was potentially preventable for the faculty and in 2 instances, 2.3%, that a death was preventable for the fellow-in-exception.

They state that these differences were not statistically significant using χ² analysis. A χ² analysis assumes that the 286 total deaths for faculty and the 88 total deaths for the fellow-in-exception are independent occurrences, and this is clearly not the case.

All observations made must be viewed as correlated data, which unfortunately violates the independent assumption for χ² testing. Should some other statistical analysis have been used? Provider-specific errors analysis resulted in counts that were very small. Again this calls into question why they chose the χ² analysis.

Also, past fellows were surveyed in another part of the study using a questionnaire that evaluated perception of training experience using a six-point Likert. Why was a paired T test done, which assumes approximately normally distributed differences, instead of a nonperimetric test?

Now having said all of that, even with these concerns about the statistical analysis chosen, I do feel that the data provide some support for the authors’ claim that patient outcomes are essentially unchanged when comparing the 2 groups. However, I have 1 last question: What conclusion if any can be generalized beyond the University of Penn in an excellent trauma program?

Over a decade ago I, along with Lew Flint and Lynn Jacobs, published in The Journal of Trauma an article on the education of the trauma surgeon in the 21st century. We knew then and we know now that the real challenge is being able to performance of outcome assessment. I have to commend the authors for taking the lead in this analysis. Thank you very much.

Dr. C. William Schwab (Philadelphia, Pennsylvania): Let me take the second question first and say: Is this generalizable to other training programs? I don’t know if it is. We haven’t been able to translate it to another trauma fellowship training program. We currently use it in another level I trauma center that we staff outside of Philadelphia which involves residents. It is interesting to see that with residents this near concurrent review and provider-specific database allows to us make some observations about their performance.

As to the statistical questions, I could proclaim to be knowledgeable about the statistical model used, but in fact, the manuscript recognized that we sought outside consultation of the type of testing we should use. So therefore, Dr. Britt, I can’t go explain why these tests were used. Based on your concerns, we will revisit with our consultant and explore your questions. Thank you.

Dr. Anthony A. Meyer (Chapel Hill, North Carolina): Dr. Schwab, obviously your program is very successful. I just was curious in terms of your structure of comparing both the requirements for surgical critical care training from the American Board of Surgery and then the structure of you having independent practitioners or semi-independent practitioners in their second year.

Are you able to cover that in the potential exception of having the 1 year of clinical experience in the ICU in the 2 years? And then the corollary to that, how do you fund that or how do you cover that since if somebody is in an RRC-approved training program they can’t bill for any of their clinical activity? Thanks.

Dr. C. William Schwab (Philadelphia, Pennsylvania): They are excellent questions.

First of all, let me say that the 12 months of surgical critical care, which is divided predominately from adult to pediatric, are supervised. The fellows literally go from two-to-three month blocks in trauma to critical care and vice versa. So 2 months in trauma, then the next 3 months is critical care.

The way we fund each segment is totally different. The surgical critical care training program is funded as an RRC-approved training slot. When they function in the 7 months in the second year as trauma fellows-in-exception, in the eyes of the State of Pennsylvania and our hospital, they are attending of staff surgeons. That is funded from clinical revenue that is generated by our group practice.

The surgical critical are fellows in all aspects of training are supervised 24/7. It is only in the 7 months of the second year when the fellow-in-exception can stand independent call and practice independently.