If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Besides quantitative results, qualitative aspects of surgical learning are vital.
•
Factors experienced as positive contributions to learning should be taken in mind.
•
Mutual learning contributes to team learning.
•
Shared goals in the OR contribute to cognitive embedding.
Background
When introducing new techniques, attention must be paid to learning curve. Besides quantitative outcomes, qualitative factors of influence should be taken into consideration. This retrospective cohort study describes the quantitative learning curve of complex endovascular aortic repair (EVAR) in a nonhigh-volume academic center and provides qualitative factors that were perceived as contributors to this learning curve. With these factors, we aim to aid in future implementation of new techniques.
Methods
All patients undergoing complex EVAR in the Leiden University Medical Center (LUMC) between July 2013 and April 2021 were included (n = 90). Quantitative outcomes were as follows: operating time, blood loss, volume of contrast, hospital stay, major adverse events (MAE), 30-day mortality, and complexity. Patients were divided into 3 temporal groups (n = 30) for dichotomous outcomes. Regression plots were used for continuous outcomes. In 2017, the treatment team was interviewed by an external researcher. These interviews were reanalyzed for factors that contributed to successful implementation.
Results
Length of hospital stay (P = 0.008) and operating time (P = 0.010) decreased significantly over time. Fewer cardiac complications occurred in the third group (3: 0% vs. 2: 17% vs. 1: 17%, P = 0.042). There was a trend of increasing complexity (P = 0.076) and number of fenestrations (P = 0.060). No significant changes occurred in MAE and 30-day mortality. Qualitative factors that, according to the interviewees, positively influenced the learning curve were as follows: communication, mutual trust, a shared sense of responsibility and collective goals, clear authoritative structures, mutual learning, and team capabilities.
Conclusions
In addition to factors previously identified in the literature, new learning curve factors were found (mutual learning and shared goals in the operating room (OR)) that should be taken into account when implementing new techniques.
Introduction
Originally introduced in aircraft manufacturing, learning curve studies addressed the variation in costs and labor time, when production quantity increased.
A learning curve can provide information on different levels: whether learning took place, at which rate, whether a desired level of performance was reached, and whether learning has stopped or even regressed. A classic learning curve (Fig. 1) is comprised of 3 phases; an initial gentle slope, followed by a phase of rapid learning, ending in a plateau phase when additional procedures no longer improve performance. Reaching the plateau phase can either mean learning has stopped and adjustments should be made in order to improve again, or that an expert plateau phase or target values are reached. More recently, decline has been added as a potential fourth phase. Here, competence decreases due to an increase in challenging cases or due to ‘unlearning’.
In order to optimize implementation of new surgical techniques, it is important to examine factors that contribute to a successful learning curve. Several technical factors have been established, such as a surgeon's manual dexterity and the experience of the supporting surgical team.
Besides technical skills, deliberate team selection and a shared mental model of motivation are vital. In addition, a balance should be sought between authoritative leadership and a safe environment, in which, all team members feel free to communicate their thoughts.
Another important factor is team stability, which enhances relational competence, knowledge of individual team members' preferences, and the way team roles relate to each other. However, when procedures become too much of a routine, adapting to change becomes difficult. This can be addressed by performing trials of new routines.
Establishing a learning curve is common practice in robotic and minimally invasive surgery. Switching from an established standard to a technically challenging new approach requires justification and is therefore particularly suited for learning curve analysis.
In the field of vascular surgery, complex endovascular aortic repair (complex EVAR) represents such a development. Complex aortic aneurysms extend up to or above the renal arteries, involving visceral and arch branches that need to be incorporated in the reconstruction. For decades, open reconstruction was the standard of treatment, albeit associated with high morbidity and mortality.
Fenestrated endovascular aneurysm repair is associated with lower perioperative morbidity and mortality compared with open repair for complex abdominal aortic aneurysms.
Complex EVAR is technically more demanding than conventional EVAR; stent grafts are tailor-made for each patient, implantation is supported by advanced imaging tools, high-end operating facilities are necessary, and it requires a treatment team to adopt new skills. These complex techniques were pioneered in high-volume aortic centers of excellence and the most robust outcome data derives from their results.
Complex endovascular aneurysm repair is associated with higher perioperative mortality but not late mortality compared with infrarenal endovascular aneurysm repair among octogenarians.
Previously, learning curves were established for branched-fenestrated EVAR implementation in the Unites States on the experience of a single surgeon and the usage of a specific device.
These studies focused on the quantitative aspect, as is often the case for surgical learning curves. The current study focuses on the qualitative aspect of implementing complex EVAR and presents quantitative outcomes in a nonhigh-volume hospital. By establishing factors that positively contributed to team learning, we aim to support the future implementation of new techniques, not just in endovascular surgery but in other surgical fields as well.
Methods
Complex EVAR Implementation
In July 2013, the first complex EVAR procedure was performed at the Leiden University Medical Center (LUMC), a tertiary referral center for aortic pathology. To optimize the implementation of complex EVAR, a dedicated endovascular treatment team (ETT) was formed, consisting of vascular surgeons, interventional radiologists, thoracic surgeons, anesthesiologists, clinical neurophysiologists, radiology technicians, and scrub nurses. Great attention was paid to team composition; members were selected based on self-professed interest, capabilities, and time commitment. All interventional radiologists and vascular surgeons had previous experience with conventional EVAR.
Prior to surgery, each patient was discussed in a multidisciplinary ETT consultation. Complex EVAR stents were designed for each individual patient based on CT imaging by the interventional radiologists, vascular surgeons, and stent graft manufacturers. The first 4 procedures were proctored. Postoperative care was planned by consulting selected intensive care unit [ICU] specialists and internal medicine doctors. 7 of the initial 20 members left the team due to job changes or retirement: a scrub nurse (2019), radiology assistant (2018), industry representative (2017), 2 thoracic surgeons (2019, 2018), anesthesiologist (2018), radiology product expert (2017). Eight members joined the team: a scrub nurse (2019), radiology assistant (2018), industry technician (2017), thoracic surgeon (2019), 2 anesthesiologists (2018, 2017), radiology product expert (2017), and a vascular surgeon (2020).
Data Collection
A single-center retrospective study was performed. All patients who had undergone complex (thoraco) abdominal EVAR in the LUMC between July 2013 and April 2021 were included. Solitary thoracic EVAR (TEVAR) procedures were excluded. Patients had followed standard of care follow-up, in accordance with our institution's protocol. They were seen in the outpatient clinic by the vascular surgeon at 6 weeks, 6 months, 12 months, and yearly after that. CT angiography, duplex ultrasonography, and abdominal X-ray were used to monitor aneurysm or stent graft-related complications. For the quantitative analysis, data was subtracted form patient's medical record, and stored in a secured computerized database. Data collection was approved by the institution's Medical Ethics Committee (METC).
In 2017, when 46 complex EVAR procedures had been performed, semistructured face-to-face interviews were conducted with all 19 ETT members by an external interviewer, in order to monitor the implementation phase.
The main goal was to identify what every team member needed in order to adequately fulfill their task. These interviews were re-examined for the qualitative analysis of the current study. All interviews were conducted within an 11-day period during which no complex EVAR procedures were conducted, in order to avoid recency bias between the interviewees. Each interview lasted between 50 and 100 min. Questions are added in Appendix A. The interviews were transcribed and coded using Atlas.ti software. ETT members’ reflections were captured in first order codes that closely followed the phrasing used by interviewees. Factors were considered to be vital to the procedure if they were mentioned at least 10 times by at least 5 different interviewees. First order codes were subsequently grouped into second order codes by the external researcher. In a final step, the researcher aggregated the second order codes into 3 key dimensions as follows: relational embedding, cognitive embedding, and team learning. Factors that influenced the learning curve according to interviewees were extracted from this data for the purpose of the current study. They were compared with factors derived from literature. Corresponding and supplementary factors are discussed in this paper.
Outcome Measures
Surgical process outcomes were initial technical success (achieved if all arteries were successfully treated as planned), operating time (minutes), blood loss (milliliters), fluoroscopy time (minutes), and volume of contrast (milliliters). Clinical patient outcomes were length of hospital stay (days), discharge to a rehabilitation center, 30-day mortality, major adverse events (MAE; complications with a Clavien-Dindo score of III-IV), the necessity of endoleak repair, and reinterventions due to complications.
In order to monitor changes in complexity over the years, the ETT established a complexity coding scheme. Complexity levels were defined as 1 (least complex), 2, 3, and 4 (most complex). Level 1 included complex EVAR with 1 or 2 fenestrations. Level 2 included 3 or 4 FEVER. Level 3 included all BEVAR patients, and level 4 included branched-fenestrated EVAR combinations, arch EVAR, and emergency cases.
Comparison of outcomes for double fenestrated endovascular aneurysm repair versus triple or quadruple fenestrated endovascular aneurysm repair in the treatment of complex abdominal aortic aneurysms.
Patients were divided into 3 temporal groups as follows: the first 30 patients (group 1), the second 30 patients (group 2), and the third 30 patients (group 3). These cut-off points are in accordance with previous research.
They were set before any analyses were made, in order to preclude bias resulting from data-dependent splitting. Baseline characteristics and outcomes are presented as numbers and percentages for categorical data and as mean or median, with standard deviation or interquartile range respectively, for continuous data. Baseline characteristics were compared using the ANOVA F-test for continuous normally distributed data. The Fisher's exact test was used for dichotomous baseline data and categorical learning curve outcomes. In addition, Poisson regression analyses were made. The learning curve for continuous outcomes was established by calculating the regression coefficients, as this is the preferred statistical method.
A multivariate regression analysis was performed to determine the effect of complexity on these continuous outcomes. In all analyses, a P-value below 0.05 was considered to indicate a statistically significant difference. All analyses were conducted using IBM SPSS Statistics version 27.
Between July 2013 and April 2021, 90 patients with complex aortic aneurysms were treated. Figure 2 shows how many patients were treated each year. A steady increase occurred in the first 4 years. Table I shows the baseline characteristics of these patients: 74 (82%) were male and mean age was 73.6 years (SD = 6.3). Columns 3–6 of Table I show the baseline characteristics of the 3 temporal groups, with P-values indicated. The groups were comparable on all variables, including age, gender, body mass index [BMI], comorbidities, risk factors, and American Association of Anesthesiologists [ASA]-score. No statistical differences were detected between baseline characteristics.
Fig. 2Number of complex EVAR procedures per year. EVAR = endovascular aortic repair.
Factors were considered to be vital to the procedure if they were mentioned at least 10 times by at least 5 different interviewees. They were divided into factors enabling relational embeddedness, cognitive embeddedness, and team learning. One factor enabling relational embedding was adequate communication. Communication should occur on a frequent basis, among all members of the ETT and in formal as well as informal settings. In the preparatory phase, precase multidisciplinary briefings provided formal occasions of communication. During surgery, “thinking out loud” by the performing surgeons and interventional radiologists enabled involvement of all participating team members in the OR. In addition to communication during official meetings and in the OR, informal discussion and social gatherings provided occasions of valued interaction. Communication was supported by a culture of mutual trust; all team members felt free to openly share their opinions and raise concerns. This depended on the team environment, which was created over time.
According to the interviewees, a shared understanding of different team roles was vital for successful team performance. This contributed to cognitive embedding. Differences in hierarchical positions and the line of command between team members were accepted by all team members. Clear authoritative structures had to be present and unquestioned, while at the same time maintaining mutual trust. Another contributing factor to cognitive embedding was a strong sense of shared responsibility and collective goals. This included attendance of all precase and postoperative meetings. It also encompassed a realization of the interdependence between team members. Due to the necessity of each team member's contribution, participants should be able to rely on each other, and therefore feel the obligation to enable the task execution of others. We found that the importance of this factor extended into the OR. During the procedure, it was expected that all conversations only concerned the treatment being performed, and all members focused on their task, while being dedicated to the overall team performance. The feeling of responsibility exceeded planned working hours.
Three factors were identified as having contributed to team learning. Team members shared their knowledge, which meant that more information that was strictly necessary to perform an assigned task was exchanged. Again, debriefing was important in this matter. In addition, acquired skills and experiences were shared for other team members to learn from them, along with relevant developments in the different disciplines involved: mutual learning. Besides becoming familiar with the technical aspect of the procedures, team members also had to become acquainted with the other members' way of work and preferences. Familiarity with each other's body language and specific preferences contributed to successful task performance and built team capabilities.
Quantitative Assessment
Figure 3A–D show the quantitative learning curves for blood loss, operating time, length of hospital stay, and volume of contrast. Specifics are depicted in Table II. It shows a statistically significant decline in operating time (Time (minutes) = 361.1–1.235 x number of procedures, P = 0.010, CI: −2.17;-0.30) and length of hospital stay (length of stay (days) = 14.7–0.102 x number of procedures, P = 0.008, CI:-0.18;-0.03). No statistically significant trend was detected for blood loss (blood loss (milliliters) = 1,433.7–3.394 x number of procedures, P = 0.345, CI: 10.50; 3.71) or contrast use (volume of contrast (milliliters) = 180.8 + 0.240 x number of procedures, P = 0.480, CI: 0.43; 0.91). No changes in statistical significance occurred after correction for complexity level. The adjusted regression coefficients were −1.344 (CI: −2.25;-0.44, P = 0.004) for operating time, −5.665 (CI: −12.43; 1.10, P = 0.100) for blood loss, 0.365 (CI: −0.33; 1.06, P = 0.300) for volume of contrast, and −0.106 (CI: −0.17;-0.04, P = 0.003) for length of stay.
Fig. 3Regression plots showing a statistically significant decline in (A) length of hospital stay (P = 0.008) and (B) operating time (P = 0.010). No significant trends were detected for (C) contrast use (P = 0.480) or (D) blood loss (P = 0.345).ml = milliliters.
Table II shows that there was no significant difference between the 3 temporal groups in 30-day mortality (1: 3%, 2: 10%, 3: 7%, P = 0.435), MAE's (1: 17%, 2: 30%, 3: 27%, P = 0.554), initial technical success of the procedure (1: 93%, 2: 90%, 3: 83%, P = 0.592), and freedom from endoleak repair (1: 50%, 2: 33%, 3: 77%, P = 0.081), or freedom from reinterventions for stent graft or aneurysm complications (1: 67%, 2: 73%, 3: 53%, P = 0.612). The number of patients discharged to a rehabilitation center did not significantly differ between the groups of experience (1: 20% vs. 2: 20% vs. 3: 13%, P = 0.964). The Poisson regression coefficients were 0.009 for 30-day mortality (P = 0.572), 0.006 for MAE's (P = 0.497), −0.002 for initial technical success (P = 0.721), −0.011 for endoleak repair (P = 0.257), −0.021 for other reinterventions (P = 0.107), and −0.001 for discharge to a rehabilitation center (P = 0.874). This indicates, on a log scale, no statistically significant differences in the expected changes for these outcomes, when the number of treated patients increases.
Figure 4A, B show the trends in postoperative complications per temporal group, presented by the type of complications (Fig. 4A) and by Clavien–Dindo score (Fig. 4B). It shows a significant decrease in the percentage of patients with cardiac complications in group 3 vs. group 2 and 1. (3: 0%, 2: 17%, and 1: 17%, P = 0.042). In addition, it shows an increase in the percentage of patients with access complications, but this ascending trend was not statistically significant (P = 0.322). There were no significant changes in the severity of complications.
Fig. 4(A) Percentage of patients with one or more postoperative complications with a Clavien–Dindo score of I-IVB, presented per group of experience (n = 30). The decline in cardiac complications was statistically significant (P = 0.042). (B) Percentage of patients with one or more postoperative cardiac, renal, access, wound, pulmonal, intestinal, or spinal complications per group of experience, presented by Clavien–Dindo score. There were no statistically significant differences between groups.
There was a trend toward an increase in complexity over the years. Table I shows that the number of procedures with a complexity score of 3 is larger in group 3 compared to group 1 (n = 18 vs. n = 8), which was mainly caused by an increase in FEVAR procedures with 4 fenestrations (n = 10 vs. n = 3). The number of procedures with level 1 complexity is smaller in group 3 compared to group 1 (n = 2 vs. n = 7). However, these trends did not present a statistically significant difference in overall complexity scores between groups (P = 0.076). There was also a trend toward an increase in the number of fenestrations per procedure, mainly caused by an increase in procedures with 3 fenestrations and a scallop, and procedures with 4 fenestrations. This trend was not statistically significant (P = 0.060).
Discussion
Qualitative Analysis
According to our interview data, factors thought by the interviewees to have positively influenced the learning curve were: communication, mutual trust, a shared sense of responsibility and collective goals, clear authoritative structures, knowledge sharing, mutual learning, and team capabilities. When implementing new techniques in the future, several of these factors can be encouraged from the start, for example by organizing team meetings in which experiences are shared and mutual learning is supported.
Figure 5 (supplementary material) shows a comparison between our factors of influence and corresponding literature factors.
The need for a shared sense of responsibility and collective goals resembles the ”shared mental model” introduced by Aveling et al.: team members should be motivated, focused, and dedicated.
Our analysis complements this factor by adding that this attitude should extend into the OR. Parker et al. discusses surgical leadership, which our factor of clear authoritative structures partially encompasses.
A contributing factor revealed by our analysis but not discussed in the literature concerning learning curves, is mutual learning. Experiences and relevant developments in the different disciplines should be shared with other team members, even if this strictly extends beyond the scope of their assigned tasks.
Fig. 5Supplementary material: Factors influencing the learning curve identified in the interviews and corresponding factors deriving from literature (in cursive).
This might be absent in our findings because additional techniques are introduced on a rolling basis in complex EVAR. Moreover, each procedure is adjusted to the specific configuration of the aneurysm that is being treated, which prevents complex EVAR from becoming routine surgery.
Quantitative Analysis
Despite a trend of increased complexity, operating time declined, which indicates that technical learning took place. This increase in complexity is in line with existing literature that shows that with growing exposure, complex EVAR became technically more demanding.
Length of hospital stay, with a cluster of patients with a long length of stay in the earlier treatment stage (Fig. 2A), significantly decreased as well.
Despite a slight trend of increased 30-day mortality and major complications (mainly between groups 1 and 2), these results were not statistically significant and adverse outcomes remained at comparable levels. In addition, the number of cardiac complications declined. Our 30-day mortality and MAE numbers can be compared to previous research with a higher number of patients, such that by Oderich et al. (30-day mortality of 1.8–8.2%, MAE of 32–36%, depending on complexity) and Tran et al. (30-day mortality of 8.6%, MAE of 21.1–23.5%, depending on complexity).
Complex endovascular aneurysm repair is associated with higher perioperative mortality but not late mortality compared with infrarenal endovascular aneurysm repair among octogenarians.
In addition, Mirza et al. presented a decline in 30-day mortality and the incidence of MAE, although the incidence of MAE in our third group of experience (27%) resembles the incidence in the final group in Mirza et al. (29%). It should be noted that these studies described specific subgroups, whereas the current study included all types of complex EVAR. This impedes meaningful comparisons of quantitative results.
Possible Confounding Factors
When interpreting a learning curve, confounders need to be taken into account. Procedural changes and adjustments in patient selection did occur during 8 years of treatment. Although a change in patient characteristics was not identified, in our experience more complicated cases were taken on based on combinations of case complexity and aneurysm configurations. Our complexity score was solely based on stent graft configuration, which does not include all aspects of a procedure's difficulty level. Factors that could be included for a more comprehensive complexity score are tortuosity, the way access was gained, and whether target vessel stenosis was present. Although consistency was aspired within the ETT, changes in team composition did occur over the years, as specified in section 2.1.
Another possible confounder is the fact that new innovations were introduced in the endovascular program, such as carbon dioxide flushing of thoracic stents to prevent cerebral air embolisms, and branched and fenestrated arch-EVAR. The slight rise in access complications could be due to the introduction of percutaneous femoral access.
Strengths and Limitations
The current study consecutively included all patients who underwent complex EVAR during 8 years of treatment. Because treatment took place in a single center, inclusion was limited to 90 patients. This represents an unselected “real world” complex EVAR population, and provides insight in the outcome numbers of a nonhigh-volume center. However, it does limit the extent of the analyses. If more patients were to be included, additional analyses could be performed, such as multivariate learning curve analysis corrected for confounding factors. With the current sample size, this could not be performed in a robust fashion. The temporal groups enabled us to determine whether learning took place. The regression plots depict the rate of learning and indicate that learning has not stopped. Although widely used and accepted, our methods only partially describe the shape of the underlying learning curve (Fig. 1). Future research aims to establish a mathematically more rigorous approach. This would provide a more thorough comparison of different learning curves and could enable treatment teams to determine what stage of learning they are in.
Another challenge is the fact that the qualitative data were gathered inductively, without ex ante referencing the medical learning curve literature. Suggestions regarding the interaction between qualitative and quantitative results thus depend on our interpretation. In future research, quantitative and qualitative data should preferably be examined prospectively. This enables researchers to investigate whether findings in the quantitative curves are reflected in the interviews, and vice versa. However, this fundamentally contradicts the inductive approach we took, which has as a core strength that interviewees dictate which factors are most important.
Conclusion
This study presented a quantitative as well as a qualitative analysis of the complex EVAR learning curve in a non-high-volume hospital. Despite a trend of increased complexity, operating time, length of hospital stay, and cardiac complications declined. Thirty-day mortality and MAE showed no statistically significant changes. We found that several factors that were previously identified in other fields extend to the field of complex EVAR: adequate communication, a shared sense of responsibility, mutual trust, clear authoritative structures, and team capabilities. The factors mutual learning and shared goals during treatment in the OR were added by our research and can aid in the future implementation of new techniques. With complexity bound to increase, monitoring progress and striving for optimization of team learning will become even more relevant.
Fenestrated endovascular aneurysm repair is associated with lower perioperative morbidity and mortality compared with open repair for complex abdominal aortic aneurysms.
Complex endovascular aneurysm repair is associated with higher perioperative mortality but not late mortality compared with infrarenal endovascular aneurysm repair among octogenarians.
Comparison of outcomes for double fenestrated endovascular aneurysm repair versus triple or quadruple fenestrated endovascular aneurysm repair in the treatment of complex abdominal aortic aneurysms.
00057-2&id=gr1.jpg){kind=link}
00057-2&id=gr2.jpg){kind=link}
00057-2&id=gr3.jpg){kind=link}
00057-2&id=gr4.jpg){kind=link}
00057-2&id=gr5.jpg){kind=link}