The influence of familiarity between the surgeon and their assistant on patient outcomes: a prospective observational cohort study

Abstract

Background:

The inverse relationship between increased surgical team familiarity and reduced operative time is established, but its effect on patient outcomes remains uncertain.

Materials and methods:

A prospective cohort study including operations by attending surgeons between 1 November 2020 and 31 December 2021 across fourteen surgical departments from four French university hospitals. Surgical team familiarity was measured as the cumulative number of previous operations performed by the same dyad of attending and assisting surgeons. Composite of adverse events within 30 days of surgery encompassed major surgical complication, unplanned reoperation, extended ICU stay, and death. We used multivariable generally estimated equations to model the association between patient outcomes and surgical team familiarity, using a logarithmic function. The model considered the clustering of operations within surgeons.

Results:

Our analysis included 8546 operations by 1109 surgical team dyads, involving 45 attending surgeons and 369 assisting surgeons. We observed a significant inverse association between surgical team familiarity and composite adverse events odds ratio [OR] 0.92 (95% confidence interval [95% CI] 0.87–0.98), major surgical complications OR 0.93 (95% CI 0.88–0.99), and unplanned reoperations OR 0.88 (95% CI 0.78–0.99), with non-significant trends observed for extended ICU stays OR 0.88 (95% CI 0.75–1.04) and deaths OR 0.87 (95% CI 0.74–1.03). Within the first 15 collaborations, this was illustrated by a reduction in the occurrence of composite adverse events from 23.0% (95% CI 22.1%–24.0%) to 16.5% (95% CI 14.1%–18.8%), major surgical complications from 21.3% (95% CI 20.3%–22.2%) to 15.3% (95% CI 13.0%–17.5%), unplanned reoperations from 8.8% (95% CI 8.6%–9.1%) to 5.2% (95%CI 4.2%–6.1%), extended ICU stays from 4.3% (95% CI 4.1%–4.5%) to 3.1% (95% CI 2.0%–4.1%), and deaths from 2.3% (95% CI 2.1%–2.5%) to 1.4% (95% CI 0.9%–1.8%).

Conclusions and relevance:

This study emphasizes that heightened familiarity among surgical teams is associated with a significant reduction in major adverse events. Building stable operating room teams should be a management priority to enhance patient outcomes.

Introduction

Surgery is a complex process with high risk for patients, requiring trust and seamless coordination among surgeons, anesthesiologists, and nurses to reliably accomplish collective tasks. Collaborative effort between team members is essential for optimal performance in the operating room. Well-trained teams provide safer patient care^[1], while poorly functioning teams may increase the risk of error and jeopardize patient outcomes^[2,3]. Successful teamwork involves a range of non-technical skills among surgical staff, such as team synergy, interpersonal communication, and situational awareness^[4]. Developing these habits of working together can enable staff to share a mental plan of the procedure, anticipate and communicate safety issues in real-time, and prevent their occurrence or address them early on^[5,6].

The concept of team familiarity reflects shared history and growing experience over time among operating room members in conducting routine surgery. As a marker of stability, it can be quantified by their number of previous collaborations. Accumulated evidence demonstrates that increasing familiarity among operating room teams is associated with shorter operative times, a relationship now well-established in various surgical specialties^[7–9]. In particular, stable dyads of attending and assisting surgeons have been shown to reduce operating times for total knee replacement, mammoplasty, and cholecystectomy^[10–12]. Furthermore, the influence of attending surgeon and anesthesiologist dyads on patient outcomes was recently observed in complex digestive and cardiac surgery^[13–15]. However, evidence is lacking to support the notion that increasing team familiarity between attending and assisting surgeons reduces patient complications. Previous studies have yielded heterogeneous findings from limited samples, offering unclear and non-generalizable guidance due to their focus on specific procedures in narrow care settings^[7].

This study was primarily designed to address this knowledge gap and determine, across various specialties, whether previous collaborations within surgical teams influenced the occurrence of major surgical adverse events. We hypothesized that increasing familiarity between attending and assisting surgeons would lead to better patient outcomes. Additionally, we were interested in examining the shape of their teaming curve to identify when they would provide optimal performance after reaching a critical threshold of collaborations.

Methods

Study design, population, and exposure

We conducted a prospective observational cohort study across 14 surgical departments from four university hospitals in France specialized in digestive, orthopedics, gynecology, urology, cardiac, thoracic, and endocrine surgery. A cohort of 45 attending surgeons, each performing a minimum of 50 procedures per year, was formed. All surgical procedures performed between 1 November 2020 and 31 December 2021 were considered for inclusion in analysis. We excluded operations for patients under 18 years old or those who had refused to share their personal data, as well as operations for palliative care or organ donation, and operations with missing operative timestamping or critical data for case-mix adjustment. Additionally, operations performed by a single attending surgeon without an assisting surgeon were not analyzed. The operations performed by surgical team dyads exceeding 50 previous collaborations were ultimately not included in final analysis due to the limited sample size, which would not allow for accurate estimations.

We estimated the association between patient outcomes and their exposure to surgical team familiarity. For each operation performed by the cohort of surgeons during the study period, surgical team familiarity was measured based on the cumulative number of previous collaborations between the attending surgeon and the assisting surgeon. These previous collaborations corresponded to every operation performed together by those distinct dyads of attending and assisting surgeons.

According to European General Data Protection Regulation No. 2016/679, this observational study was based on anonymous data and approved by the French Committee of Expertise in Research, Studies, and Evaluations in the Health domain, the French National Data Protection, and the European Research Council Executive Agency. Additionally, this study was deemed exempt from formal oversight by the Mass General Brigham Institutional Review Board. Patients were informed that their health data might be reused for research purposes and that they had the opportunity to opt out. Each surgeon in the cohort provided informed consent to participate in the research and for the use of their data.

Data sources and outcomes

For each operation, data were prospectively collected from a homogeneous information system across the Lyon University hospitals, encompassing details regarding the surgical procedure, patient, surgeons, and operating rooms involved. Specifically, we gathered data on the patients’ socio-demographic characteristics (age, gender, social coverage, and median income), care provided during hospital stays, including details about the type of surgery performed (organ, surgical approach, and complexity), and the primary diagnosis linked to the operative indication.

We supplemented this information with data collected by clinical research assistants from the patients’ electronic health records to acquire details such as the scheduling of the operation (elective, semi-urgent, and urgent), occurrences of intraoperative and postoperative adverse events, and the preoperative comorbidities of the patient (see Appendix Methods S1 for comorbidities list). We also extracted data from the anesthesia records to identify the type of anesthesia administered and the patients’ ASA Physical Status Classification System.

In addition, for each operation, we used data from the operating room management software and the surgical procedure report to determine the scheduling of each operation with the attending and assisting surgeons primarily involved. Lastly, we gathered human resources data to determine the age and professional status (professor versus non-professor) of the attending surgeon and the experience of the assisting surgeon (junior versus senior).

Inspired by the Clavien–Dindo classification, which considers surgical major adverse events in an objective and reproducible manner^[16], the primary outcome was a composite measure encompassing adverse events during surgery or within the 30-day post-surgery period. These events included major intraoperative or postoperative complications (see Appendix Methods S1 for complications list across various categories: general, infectious, hemorrhagic, parietal, cardiopulmonary, neurological, abdominopelvic, orthopedic, cervical, and functional), unplanned reoperation for complication arising from the initial surgery, extended stays postoperatively in the intensive care unit (ICU) for at least two nights or in the intermediate care unit (IMCU) for at least five nights due to organ failure, and intraoperative or postoperative mortality. For secondary outcomes, we considered each event separately.

Statistical analysis

Operation characteristics were presented using the mean and standard deviation for continuous variables and frequencies with percentages for qualitative variables, while attending and assisting surgeons’ characteristics were presented using the median and interquartile range (IQR) for continuous variables. For each outcome, we compared the average team familiarity, based on the number of previous collaborations between surgeons, with respect to the occurrence of adverse events, and by considering the clustering of operations within attending surgeons.

We then employed multivariable generalized estimating equations (GEEs) with an independent correlation matrix to model the association between patient outcomes and surgical team familiarity. A spline analysis was performed, and we tested various shapes in this relationship, showing that a logarithmic association with the number of previous collaborations of the surgical team dyads provided the best model fit. In the final GEE, we included as fixed effects the age and professional status of the attending surgeon, the experience of the assisting surgeon, and the patient’s preoperative risk score categorized into quartiles and surgical specialties. The attending surgeon identity was considered in the models as a random effect, and possible interactions between the covariates were examined. It is important to note that preoperative risk scores were created using outcome prediction models developed from an independent dataset of patients operated by the same cohort of surgeons between 1 January 2022 and 31 October 2022. Potential confounders considered for case-mix adjustment included patient age, sex, comorbidities, American Society of Anesthesiology (ASA) score, socioeconomic status, surgical specialty, surgery indication and approach, procedure complexity and scheduling, and the type of anesthesia (see Appendix Methods S1 for details).

To test the robustness of the study’s main findings, we conducted a sensitivity analysis by excluding operations performed when the assisting surgeon was considered senior in the surgical team dyad. Thus, only surgical residents were retained in this analysis as assisting surgeons, considering that their rotations traditionally start in November in France, synchronously with the study period. In this case, we also added the age of the assisting surgeon to the model.

Given that our study is observational, the necessity of adjusting for multiple comparisons when working with real-world data, as opposed to random numbers, is debatable for ensuring correct interpretation^[17]. However, we adhered to a predefined protocol that identified the composite adverse event as the primary endpoint^[18] and employed a fixed-sequence testing approach, following an order of increasing severity for secondary endpoints – major surgical complication, unplanned reoperation, extended ICU stay, and death – with a significance threshold of P < 0.05^[19]. This structured approach minimizes the risk of spurious findings and ensures correct interpretation of our results.

Data manipulation and analyses were performed using R version 4.3.0 (R Foundation, Vienna, Austria) and SAS software (version 9.4; SAS Institute Inc., Cary, NC).

We reported that our results are in line with the STROCCS criteria for observational studies^[20].

Results

Study population

Operations and surgeons’ characteristics are detailed in Table 1. The final study population comprised 8546 operations (see flow chart in Figure 1), conducted by 1109 surgical team dyads, consisting of 45 attending surgeons and 369 assisting surgeons. The median age was 44 years for attending surgeons and 30 years for assisting surgeons. Within surgical team dyads, the median number of collaborations was 25 for attending surgeons and 5 for assisting surgeons.

Table 1.

Characteristics of operations, attending surgeons, and assisting surgeons

Operations	(n = 8546)
Sex, No. (%)
Female	4889 (57.2)
Male	3657 (42.8)
Patient age, no. (%)
Mean (SD)	55.7 (17.2)
ASA class, no. (%)
1	2363 (27.7)
2	3971 (46.5)
3	2033 (23.8)
4	171 (2.0)
5	8 (0.1)
Number of comorbiditiesa
Mean (SD)	2.5 (2.1)
Patient covered by precarity social insurance, No. (%)
No	7821 (91.5)
Yes	725 (8.5)
Surgical specialty of surgery, No. (%)
Digestive	2209 (25.8)
Gynecologic	1908 (22.3)
Orthopedic	1895 (22.2)
Urologic	789 (9.2)
Cardiac	742 (8.7)
Endocrine	617 (7.2)
Thoracic	386 (4.5)
Scheduling of surgery, No. (%)
Elective	7095 (83.0)
Urgent	1218 (14.3)
Semi-urgent	233 (2.7)
Type of hospitalization, No. (%)
Inpatient	6621 (77.5)
Outpatient	1923 (22.5)
Type of anesthesia, No. (%) (missing = 2)
General	6934 (81.1)
Regional	1501 (17.6)
Local	109 (1.3)
Initial surgical approach, No. (%) (missing = 11)
Open	5086 (59.5)
Videoscopic	2181 (25.5)
Endoscopic	868 (10.2)
Robotic	400 (4.7)
Composite adverse event, No. (%)
No	6894 (80.7)
Yes	1652 (19.3)
Major surgical complication, No. (%)
No	7016 (82.1)
Yes	1530 (17.9)
Unplanned reoperation, No. (%)
No	7978 (93.4)
Yes	568 (6.6)
Extended ICU stay, No. (%)
No	8236 (96.4)
Yes	310 (3.6)
Death, No. (%)
No	8397 (98.3)
Yes	149 (1.7)
Attending surgeons	(n = 45)
Age
Median (IQR)	44.0 (39–51)
Sex, No. (%)
Man	35 (77.8)
Woman	10 (22.2)
Professional status, No. (%)
Associate/Full Professor	24 (53.3)
Non-professor	21 (46.7)
Number of surgeries
Median (IQR)	164 (117–256)
Number of previous collaborations with assisting surgeon
Median (IQR)	25.0 (15–44)
Assisting surgeons	(n = 369)
Age
Median (IQR)	30.0 (27–33)
Professional status, no. (%)
Junior	299 (81.0)
Senior	70 (19.0)
Number of previous collaborations with attending surgeon
Median (IQR)	5.00 (1–16)

^aComorbidities including critical condition, current pregnancy, obesity BMI ≥ 30 kg/m², malnourishment, tobacco addiction, alcohol addiction, other addiction, open wound, surgical site infection, sepsis, endocarditis, cancer, neoadjuvant treatment, immune deficiency, coagulopathy, anticoagulant treatment, platelet antiaggregation treatment, blood transfusion, coma, limb paralysis, other neurological disorder, confusion, dementia, depression, cardiovascular disease, neurovascular disease, peripheral arterial disease, cardiac arrhythmia, chronic heart failure, hypertension, diabetes, dyslipidemia, pulmonary artery systolic pressure > 60 mmHg, chronic renal failure, acute renal failure, chronic respiratory failure, chronic obstructive pulmonary disease, liver disease, rheumatic pathology, and hypoparathyroidism

Figure 1.

Population flow chart.

The average familiarity with the assisting surgeon varied among the 45 surgeons and across the four patient complication risk quartiles stratified by surgical specialty (P < 0.001).

The majority of patients were female (57.2%) with a mean age of 55.7 years. On average, they had 2.2 comorbidities, and 25.9% had an ASA score of 3 or higher. The operations encompassed various specialties, including digestive (25.8%), gynecologic (22.3%), orthopedic (22.2%), urologic (9.2%), cardiac (8.7%), endocrine (7.2%), and thoracic (4.5%) procedures. Most operations were elective (83.0%), performed under general anesthesia (81.1%), and involved an open surgical approach (59.5%). Composite adverse events occurred in 1652 (19.3%) operations, including 1530 (17.9%) patients experiencing major surgical complications, 568 (6.6%) undergoing an unplanned reoperation, 310 (3.6%) having a prolonged ICU stay, and 149 (1.7%) deaths.

Association between patient outcomes and surgical team familiarity

Figure 2 shows that in operations with the occurrence of composite adverse events, the surgical team dyads had a median of 5 previous collaborations, compared to 7 in operations without events (P = 0.026). Similarly, there was a median of 5 versus 7 previous collaborations in operations with major surgical complications or not (P = 0.031) and 4 versus 7 previous collaborations in operations with unplanned reoperations or not (P = 0.015).

Figure 2.

Median number of previous collaborations according to surgical outcomes.

This figure shows the median number of previous collaborations for each outcome. The statistical significance is given by an unadjusted GEE model.^*Statistically significant: composite adverse event P = 0.026, major surgical complication P = 0.031, unplanned reoperation P = 0.015, extended ICU stay P = 0.404, and death P = 0.088

Table 2 presents final multivariable models. We found a significant relationship between surgical team familiarity and composite adverse events odds ratio [OR] 0.92 (95% confidence interval [95% CI] 0.87–0.98, P = 0.007), major surgical complications OR 0.93 (95% CI, 0.88–0.99; P = 0.015), and unplanned reoperations OR 0.88 (95% CI, 0.78–0.99, P = 0.027), while a non-significant trend was observed for extended ICU stays OR 0.88 (95% CI, 0.75–1.04, P = 0.13) and deaths OR 0.87 (95% CI, 0.74–1.03, P = 0.11).

Table 2.

Team familiarity and other determinants associated with adjusted surgical outcomes

	Composite adverse event		Major surgical complication		Unplanned reoperation		Extended ICU stay		Death
Characteristic	Odds ratio (95% CI)c	P value	Odds ratio (95% CI)c	P value	Odds ratio (95% CI)c	P value	Odds ratio (95% CI)c	P value	Odds ratio (95% CI)c	P value
Surgical team familiaritya	0.92 (0.87–0.98)	0.007	0.93 (0.88–0.99)	0.015	0.88 (0.78–0.99)	0.027	0.88 (0.75–1.04)	0.132	0.87 (0.74–1.03)	0.101
Attending surgeon age	1.00 (0.99–1.02)	0.807	1.00 (0.99–1.02)	0.570	1.00 (0.98–1.02)	0.906	1.00 (0.98–1.02)	0.987	0.99 (0.96–1.02)	0.505
Attending surgeon status
Non-professor	-		-		-		-		-
Associate/full professor	0.95 (0.74–1.23)	0.717	0.95 (0.73–1.24)	0.696	0.92 (0.66–1.28)	0.615	1.07 (0.72–1.59)	0.723	0.85 (0.57–1.27)	0.435
Assisting surgeon status	-		-		-		-		-
Junior
Senior	1.35 (1.01–1.79)	0.041	1.35 (1.00–1.82)	0.049	1.22 (0.88–1.70)	0.229	1.14 (0.84–1.56)	0.393	1.25 (0.62–2.54)	0.528
Patient risk score quartile by surgical specialtyb
Cardiac and thoracic (Q1)	-		-		-		-		-
Cardiac and thoracic (Q2)	1.22 (0.64–2.29)	0.545	1.29 (0.69–2.43)	0.420	1.05 (0.31–3.57)	0.940	1.07 (0.50–2.29)	0.858	2.71 (1.19–6.16)	0.017
Cardiac and thoracic (Q3)	1.69 (1.25–2.29)	<0.001	1.59 (1.20–2.09)	0.001	2.90 (1.09–7.74)	0.033	2.69 (1.16–6.24)	0.021	6.16 (3.50–10.8)	<0.001
Cardiac and thoracic (Q4)	3.49 (2.35–5.17)	<0.001	3.21 (2.19–4.71)	<0.001	5.91 (1.94–18.0)	0.002	5.81 (2.65–12.7)	<0.001	16.4 (10.9–24.5)	<0.001
General and endocrine (Q1)	0.22 (0.15–0.32)	<0.001	0.22 (0.14–0.33)	<0.001	0.38 (0.11–1.26)	0.115	0.02 (0.00–0.13)	<0.001	0.23 (0.05–1.04)	0.056
General and endocrine (Q2)	0.56 (0.36–0.85)	0.006	0.56 (0.36–0.87)	0.010	0.64 (0.18–2.29)	0.495	0.11 (0.03–0.37)	<0.001	0.12 (0.02–0.88)	0.037
General and endocrine (Q3)	0.92 (0.62–1.36)	0.676	0.98 (0.66–1.44)	0.912	1.16 (0.37–3.63)	0.795	0.39 (0.15–1.01)	0.052	0.99 (0.44–2.26)	0.989
General and Endocrine (Q4)	3.25 (2.26–4.68)	<0.001	3.28 (2.29–4.70)	<0.001	4.02 (1.34–12.1)	0.013	2.35 (1.01–5.47)	0.048	5.24 (2.72–10.1)	<0.001
Orthopedic and urogynecologic (Q1)	0.08 (0.05–0.12)	<0.001	0.07 (0.04–0.11)	<0.001	0.13 (0.04–0.45)	0.001	0.00 (0.00–0.00)	<0.001	0.08 (0.01–0.82)	0.033
Orthopedic and urogynecologic (Q2)	0.17 (0.11–0.25)	<0.001	0.14 (0.09–0.22)	<0.001	0.38 (0.12–1.27)	0.116	0.02 (0.00–0.17)	<0.001	0.13 (0.04–0.45)	0.001
Orthopedic and urogynecologic (Q3)	0.33 (0.23–0.48)	<0.001	0.31 (0.22–0.45)	<0.001	0.62 (0.19–2.02)	0.431	0.00 (0.00–0.00)	<0.001	0.00 (0.00–0.00)	<0.001
Orthopedic and urogynecologic (Q4)	0.96 (0.64–1.44)	0.842	0.94 (0.62–1.42)	0.767	1.64 (0.52–5.15)	0.397	0.25 (0.07–0.93)	0.039	1.16 (0.51–2.63)	0.720

^aSurgical team familiarity was measured based on the number of previous collaborations between the attending and assisting surgeon and modeled using a logarithmic transformation.

^bPreoperative risk scores were developed for the five outcomes across the surgical specialties using a dataset of 3644 operations. The variables included demographics, surgical details, and comorbidities. Categories were based on quartiles (Q1-Q4) representing increasing adverse event rates by surgical specialty.

^cAbbreviations: CI = confidence interval.

We employed a multivariable generalized estimating equations (GEE) with an independent correlation matrix to model the association between patient outcomes and surgical team familiarity. The attending surgeon identity was considered in the models as a random effect, and possible interactions between the covariates were examined.

Figure 3 and Table S1. http://links.lww.com/JS9/D813 illustrate a marked reduction in the rate of composite adverse events from 23.0% (95% CI 22.1%–24.0%) to 16.5% (95%CI 14.1%–18.8%) within the first 15 collaborations. Similar patterns were observed for major surgical complications, decreasing from 21.3% (95% CI 20.3%–22.2%) to 15.3% (95% CI 13.0%–17.5%), unplanned reoperations, decreasing from 8.8% (95% CI 8.6%–9.1%) to 5.2% (95% CI 4.2%–6.1%), extended ICU stays, decreasing from 4.3% (95% CI 4.1%–4.5%) to 3.1% (95% CI 2.0%–4.1%), and deaths, decreasing from 2.3% (95% CI 2.1%–2.5%) to 1.4% (95% CI 0.9%–1.8%).

Figure 3.

Adjusted surgical outcomes probability by team familiarity.

Shaded areas represent 95% confidence intervals.The expected outcomes according to growing team familiarity were provided by a GEE adjusted for patient risk score, attending surgeon age and status, and assisting surgeon status. Surgical team familiarity was measured based on the number of previous collaborations between the attending and assisting surgeon.

These results remained robust overall in sensitivity analyses focusing on surgical team dyads with only junior assisting surgeons, as shown in Figure S1 (http://links.lww.com/JS9/D812), Tables S2 (http://links.lww.com/JS9/D814) and S3 (http://links.lww.com/JS9/D815). In the subgroup analysis of junior surgeons, we observed a non-statistically significant trend between team familiarity and patient outcomes. The direction and magnitude of the effects were consistent with our main findings, reinforcing the robustness of our primary results.

Discussion

This prospective study examined the association between surgical team familiarity, measured by the number of previous collaborations between attending and assisting surgeons, and patient outcomes across multiple specialties and operating room contexts. We observed an inverse relationship between increasing team familiarity and the decreasing occurrence of major surgical adverse events. The teaming curve followed a logarithmic shape, demonstrating significant improvement in patient outcomes over the first fifteen operations, after which performance plateaued. This suggests that the team rapidly reached a level of mature collaboration, resulting in substantial gains in patient safety. This pattern was robustly established using various outcome metrics and remained consistent when the analysis was stratified based on the experience of the assisting surgeon.

The familiarity within surgical teams has been investigated in various ways in the past. Previous studies have focused on specific surgical procedures and considered a variety of outcomes, ranging from operative time and clinical results to costs and teamwork quality. Different team dyads were examined based on the composition of the operating room staff, including surgeons, anesthesiologists, and nurses^[7,21]. In this study, our primary interest was on the surgical team dyad, as previous research consistently shows that prior collaborations between attending and assisting surgeons significantly impact surgical efficiency, though evidence regarding their effect on patient clinical outcomes remains limited^[7]. Our choice to use the number of previous collaborations as an accurate proxy for team familiarity aligns with the majority of research in this area^{[7,10,11,13,22–25]}. However, instead of focusing solely on previous collaborations for a specific surgical procedure, we considered all types of procedures that a given dyad of attending and assisting surgeons had performed together in the past. We chose this longitudinal metric to investigate surgical team dynamics over static approaches that have been erratically employed. These include classifying teams as familiar based on their previous months working together^[14], and whether the case was performed by a day team versus a night team^[26,27], by a dedicated team to a specific surgical specialty^[28], or by a standardized team^[14,29].

Our findings align with the growing evidence that familiar and stable team compositions in the operating room are associated with improved performance^[30]. Previous studies, often limited by small sample sizes and focusing on either a single specialty or individual procedure, have demonstrated that increasing team familiarity can reduce operative time, but they have not shown a significant effect on clinical outcomes^[7]. To our knowledge, a single study has suggested the potential benefit of maintaining stable surgeon pairs to reduce post-mastectomy infections requiring reoperation^[14,25]. The influence of familiarity between surgeons and anesthesiologists has also been shown to affect patient morbidity and mortality in complex cardiac and digestive surgeries^[13,15]. Our results complement these findings by illustrating the teaming curve based on previous collaborations between attending and assisting surgeons across diverse specialties, highlighting that familiar surgical teams achieve better patient outcomes.

We assume that synergistic surgical teamwork develops through familiarity, enhancing communication, mutual trust, and a shared understanding of the procedure. Familiarity allows the assisting surgeon to anticipate the attending surgeon’s preferences and gestures, facilitating task coordination and seamless, often non-verbal communication during complex procedures. Repeated collaboration within the surgical team is thought to catalyze this synergy, improving working habits and overall performance.

Confidence in our results is strengthened by several elements. Extensive data were prospectively collected from a multi-institutional cohort of surgeons through a systematic and comprehensive evaluation of patient health records by research assistants, cross-checked with other data sources to ensure validity. This approach helped avoid coding bias and limited data granularity that can result from the secondary use of larger hospital databases. Furthermore, we thoroughly adjusted for potential confounders, including operation and surgeon attributes, and we modeled surgical outcomes predictions using an independent dataset to assess preoperative patient risk. Finally, our analysis yielded robust results across various outcome metrics and remained consistent when focusing on operations performed with junior assisting surgeons only. By concentrating on the initial rotation of assisting surgeons, we captured the earliest collaborations within surgical teams, providing valuable insights into the development of team familiarity over time.

This study has important implications for scheduling team compositions in the operating room. Perioperative care systems are inherently complex and resist easy standardization and interchangeability among health professionals^[31]. The principles of Taylorism, traditionally employed in manufacturing, cannot be blindly applied to surgical services, where the uniqueness of each team member and human interactions is paramount^[32]. By revealing a substantial reduction in major surgical adverse events after only fifteen collaborations between attending and assisting surgeons, regardless of the type of surgery performed together in the past, this study supports the idea that stable operating room organization is essential for maximizing surgical team performance. Accordingly, interventions fostering team familiarity could lead to better surgical outcomes and substantial gains in patient safety. This implies the deliberate composition and maintenance of stable dyads of surgeons over time. Future research may also clarify the personal and collective mechanisms within operating room teams that explain the inverse relationship between higher familiarity and lower patient risk. A better understanding of these mechanisms is important to propose strategies to mitigate worse outcomes at lower levels of collaboration. Targeted simulation trainings or briefings with unfamiliar assisting surgeons may facilitate the sharing of a common history and safety culture^[33].

Limitations

Our work must be understood in the context of several limitations. The heterogeneous surgical population makes case-mix adjustment challenging, and residual factors may confound our results. Despite accounting for various patient, operation, and surgeon-specific variables in our adjustment scheme, it is possible that other unknown or unmeasured factors might affect the link between surgical teams’ familiarity and patient outcomes. Another limitation of our study is the assumption that the number of previous collaborations between attending and assisting surgeons was either zero or near zero at the starting date. This assumption may not fully capture the actual prior experience of some senior assisting surgeons, although we expect it to be accurate for junior assisting surgeons. Further, the allocation of assisting surgeons to attending surgeons was not random, as is typical in observational studies, limiting our ability to infer causality and instead allowing us to observe relationships between surgical team familiarity and patient outcomes. We expect minimal confounding from the junior surgeons’ learning curve for several reasons: the exclusion of procedures where attending surgeons were supervising or assisting junior surgeons; the diverse experience levels of the residents, with the majority having prior surgical experience at the start of the study; and their involvement in a wide range of procedures across various departments, which allowed them to gain familiarity with the OR environment and teamwork, rather than focusing solely on the acquisition and performance of a single task. Additionally, the study’s focus on the relationship between attending and assisting surgeons may not fully capture the complexity of interactions among all operating room personnel, as collaborations with anesthesiologists and nurses were not considered in the analysis. Lastly, the generalizability of our findings is limited by the French context within a specific geographical area.

Conclusions

In this prospective multispecialty study, increasing familiarity between the attending surgeon and the assisting surgeon was significantly associated with better patient outcomes. The results underscore the importance of fostering collaboration and stability among surgical team members as a management priority to enhance patient safety in the operating room.

TopSurgeons Study Group

We thank the members of the TopSurgeons Study Group: Jake Awtry, Lionel Badet, David W Bates, Lucie Bonin-Crepet, Olivier Cannarella, Damien Carnicelli, Martin Carrerre, Keyne Charlot, Philippe Chaudier, Gautier Chene, François Chollet, Virginie Cloud, Quentin Cordier, Ethan Cormont, Marion Cortet, Eddy Cotte, Sébastien Crouzet, Filippo Dagnino, Hassan Demian, Kim I de la Cruz, Tanujit Dey, Antoine Duclos, Xavier Dutheil, Fadi Farhat, Jean-Baptiste Fassier, Yves François, Witold Gertych, François Golfier, Romain Gorioux, Claire-Angéline Goutard, Stanislas Gunst, Muriel Hermine, Nathalie Hoen, Vahan Képénékian, Gery Lamblin, Mickaël Lesurtel, Jean-Christophe Lifante, Lucie Louboutin, Sébastien Lustig, Jean-Yves Mabrut, Laure Maillard, Jean-Michel Maury, Stéphanie Mazza, Kayvan Mohkam, Nicolas Morel-Journel, Erdogan Nohuz, Andréa Nunes, Jean-François Obadia, Léa Pascal, Arnaud Pasquer, Guillaume Passot, Elise Pelascini, Charles-André Philip, Vincent Pibarot, Stéphanie Polazzi, Gilles Poncet, Matteo Pozzi, Hugo Prieur, Maud Robert, Frédéric Rongieras, Alain Ruffion, Sophie Schlatter, Sofia Sebaoui, Elvire Servien, Sarah Skinner, Stefanie Soelling, Daniel Stelzl, Quoc-Dien Trinh, François Tronc, Delphine Vaudoyer, Laurent Villeneuve, Anthony Viste, Marco Vola, Sophie Warembourg, Joel S Weissman.

Ethical approval

In accordance with European General Data Protection Regulation No. 2016/679, this study used pseudonymized data. It was approved by the French National Data Protection Authority (DR-2020-055 CNIL) and the European Research Council Executive Agency (801660 ERCEA). Additionally, it was deemed exempt from formal oversight by the Mass General Brigham Institutional Review Board (Protocol 2023P002266). Patients were informed that their health data might be reused for research purposes and given the opportunity to opt-out. Each surgeon in the cohort provided informed consent to participate in the research and for the use of their data.

Consent

Each surgeon in the cohort provided informed consent to participate in the research and for the use of their data.

Sources of funding

This project was supported by a European Research Council (ERC) Starting Grant under the European Union’s Horizon 2020 research and innovation program (Grant agreement No. 801660 – TopSurgeons – ERC-2018-STG). This study was also supported by a public grant from the French Ministry of Health (Programme de Recherche sur la Performance du Système des Soins [PREPS-17-0008]). The funders of the study had no role in 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. The researchers confirm that they are independent from the funder of the study.

Author’s contribution

A.D. contributed to obtaining funding and supervising study. D.S. and A.D. contributed to study conception and design. D.S., S.P., J.C.L., T.D., and A.D. contributed to acquisition, analysis, or interpretation of data. S.P. and A.D. contributed to data management. D.S., T.D., and A.D. contributed to statistical analysis. D.S. and A.D. drafted the manuscript. D.S., S.P., J.C.L., T.D., and A.D. contributed to critical revision of the manuscript for important intellectual content and approved the final published version. T.D. and A.D. contributed to administrative, technical, or material support. All authors agreed to be accountable for all aspects of the work. D.S. and A.D. 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. A.D. is the guarantor, had final responsibility for the decision to submit for publication and attests that all listed authors meet authorship criteria.

Conflicts of interest disclosure

All the authors declare to have no conflicts of interest relevant to this study.

Research registration unique identifying number (UIN)

ClinicalTrials.gov ID NCT04532658.

Guarantor

Daniel Robert Stelzl and Antoine Duclos.

Provenance and peer review

Not commissioned, externally peer-reviewed.

Data availability statement

Due to confidentiality requirements, the patient and surgeon data for this project are restricted to the immediate research team.

Use of artificial intelligence

We utilized ChatGPT 4 for language revision to ensure grammatical accuracy of the manuscript. However, the content and analysis presented were independently generated and did not involve the use of any AI model.