Introduction

Determining the organizational and systems-related factors that contribute to trends in surgical outcomes is complex and often difficult. Seasonal trends have been linked to variations in outcomes, with increased rates of post-operative infections in the summer months [1], and hospital performance being affected by weekends and holidays due to a shortage of medical staff [2,3,4]. Several studies hypothesize worsened outcomes for patients receiving care at teaching hospitals in July, as new trainees enter residency training and engage in patient care responsibilities for the first time. Despite better outcomes for patients treated at teaching hospitals overall compared to those at non-teaching hospitals [5], there are reported trends of increased medical errors, hospital inefficiency, and higher morbidity to be associated with the period of resident turnover—a phenomenon popularly referred to as the July Effect [6]. During this period, new residents and trainees have been known to disrupt the multidisciplinary flow of patient care, which has been shown to rely on staff experience and a high level of physician-nurse collaboration [7]. The July Effect has been observed in numerous specialties, including neurosurgery [8], pancreatic surgery [9], and urology [10]. Data from plastic surgery and several others [8, 11,12,13,14], in contrast, demonstrate a lack of significant findings to support the presumed phenomenon in July.

At teaching hospitals, urology constitutes a large proportion of robotic-assisted surgery (RAS) cases. Moreover, RAS continues to replace conventional laparoscopy in a wide range of procedures; as such, residents are increasingly participating in numerous RAS procedures. To meet the high-volume demand of RAS, residency programs have formally begun to integrate robotic surgery training into their curriculum. From 2020 to 2021 alone, the Accreditation Council for Graduate Medical Education (ACGME) increased the minimum number of robotic cases urology residents are required to perform from 50 to 80 [15]. This increase signifies the growing importance of achieving both optimal learning curves and operating room efficiency, without compromising patient safety. The heightened involvement of urology residents in RAS provides an ideal population within which to help clarify the role of a July phenomenon in robotic surgery [16, 17]. Though RAS rapidly emerges at the forefront of minimally invasive surgery, associations between resident involvement and the July Effect in RAS have not been thoroughly investigated. Therefore, the objective of this study was to investigate the July Effect in urologic RAS at a single, high-volume academic medical center.

Materials and methods

After obtaining institutional review board (Pro00029501) approval, a retrospective chart review was conducted for robotic urological surgery cases performed at a single metropolitan tertiary medical center from March 2008 to September 2019. Patient and surgical characteristics that were collected included patient age, body mass index (BMI), time of academic year (July–October, November–February, March–June), procedure performed, any concomitant procedures, estimated blood loss (EBL), and use of da Vinci robotic console Si or Xi (Intuitive Surgical, Sunnyvale, CA). Specific operative timepoints were collected and assessed for differences: patient entrance to operating room (OR), procedure start, initial incision, incision closure, procedure stop, and patient exit from OR. Surgery time (ST) was defined as the operative duration between the initial incision and procedure stop. Non-surgery time (NT), or non-skin-to-skin OR time, was defined as the total cumulative duration from when the patient entered the OR to the initial incision, and from the procedure stop to when the patient exited the OR.

We had the unique opportunity to address the concept of the July Effect as a natural experiment due to the implementation of a urology residency at our academic center in 2011. Resident involvement in urologic cases began when our institution’s residents began working with the academic staff after the end of their internship, in July 2012. Two cohorts were constructed: cases with residents and cases without residents. After July 2012, a large proportion of cases remained uncovered by residents until a full complement of residents was present. To control for the fact that residents were introduced after an individual surgeon had amassed a certain clinical volume, we incorporated attending-specific case experience into the model.

Cases were redistributed into one of five categories to account for differences in case complexity. Category 1 (robotic-assisted laparoscopic prostatectomy (RALP)) included radical and simple prostatectomy. Category 2 (robotic-assisted sacrocolpopexy) contained only sacrocolpopexy. Category 3 (robotic-assisted cystectomy) contained only cystectomy. Category 4 (robotic-assisted nephrectomy) included radical nephrectomy, partial nephrectomy, adrenalectomy, partial retroperitoneal nephrectomy, and nephroureterectomy. Category 5 (robotic-assisted pyeloplasty) included bladder diverticulectomy, pyeloplasty, partial cystectomy, ureteral reimplantation, and ureterectomy.

Statistical analysis

A mixed effects model was developed to estimate the independent impact of each variable on ST, with each variable as a fixed effect and the particular attending surgeon as a random effect. The multivariable models used in this study factored in age, BMI, EBL, case type, da Vinci model, and whether the case involved multiple procedures. To account for significant covariates, we included a month of the year as the interaction term in the adjusted model.

Results

A total of 1728 urologic robotic cases were performed by 13 different attending surgeons during the study period. A total of 686 cases (39.7%) were performed independently by attendings, and 1042 cases (60.3%) were performed in the presence of surgical residents. Of these recorded cases, there were 1,361 (78.8%) RALPs, 124 (7.2%) sacrocolpopexies, 33 (1.9%) cystectomies, 150 (8.7%) nephrectomies, and 60 (3.5%) pyeloplasties (Table 1). Tables 2 and 3 present the differences in mean operative times by case category with and without resident presence. Average ST did not significantly differ by resident involvement at the univariate level (248.77 min without residents, 244.97 min with residents; p = 0.25) (Table 2). NT, on the other hand, significantly improved with resident presence by 4.6 min (p < 0.001) (Table 3). The unadjusted relative impact of each variable on surgery time with and without residents are presented in Table 4. Factors associated with a prolonged effect on operative time in cases with residents included increasing patient age and BMI. For each increase in BMI value, ST was estimated to increase by 1.98 min (p < 0.001) in cases with residents. The presence of a resident was associated with a slightly reduced EBL (p < 0.001). Cystectomies (Group 3) and nephrectomies (Group 4) appeared to be significantly prolonged when residents were involved.

Table 1 Recoded case categories, original cases, and observed frequencies
Full size table
Table 2 Mean operative times (in minutes) with and without resident for ST
Full size table
Table 3 Mean operative times (in minutes) with and without resident for NT
Full size table
Table 4 Multivariate analysis of unadjusted variables and their associated impact on ST for cases without resident presence and with resident presence
Full size table

Following adjustment (Table 5), resident presence was determined to be a significant independent predictor of longer ST. Overall, resident presence was associated with increased ST by 38.62 min (p < 0.001). ST in cases involving residents was shown to decrease by 8.7 min after the first 4 months of the academic year (July–October), then by an additional 5.1 min after the next 4 months (November–February) (p = 0.027, < 0.001). When the interaction term was included in the analysis, the ST of cases with residents improved by 18.94 min on average after the first 4 months of the academic year (July–October vs November–June, p = 0.001). There was significant variability to ST in cases with residents when we compared time intervals stratified by case type (Fig. 1a). Cases with attending surgeons only, in contrast, did not exhibit significant variability across case type. NT did not significantly vary across both cohorts (Fig. 1b).

Table 5 Multivariate analysis of the variables with adjusted impact on ST
Full size table
Fig. 1
figure 1

a Comparison of ST with and without resident presence by case type (created using the R Project for Statistical Computing). b NT with and without resident presence by case type (created using the R Project for Statistical Computing)

Full size image

Discussion

This study investigated the impact of resident presence on robotic operative times in urology at a single high-volume academic hospital. We found a July Effect on operative times for urologic robotic surgery (U-RAS), as ST with residents were significantly longer between July to October. This finding expands on results from other studies that investigated the effects of resident involvement on operative efficiency throughout different periods of the academic year. In cardiac surgery, operative times averaged 295 ± 90 min (p < 0.05) in surgeries performed from July to August. The rest of the academic year (September to June) demonstrated an improvement with a reduced operative time of 288 ± 90 min (p < 0.05) [18]. These findings are consistent with other studies investigating the impact of resident training in urology demonstrating increased operative times and no difference in overall complication rates [10, 19,20,21]. The impact of resident training on minimally invasive surgery outcomes, however, remains largely unexplored.

The natural dichotomization of cases without (prior to July 2012) and with resident trainees (July 2012 and after) allowed us to explore the impact of resident presence on operative times for urologic cases. Most previous studies that investigated the July Effect did not have this advantage, thus largely assuming a causative relationship between trainee involvement and prolonged operative times [8, 18]. The study by Afshar et al. [10] is notable as their analyses of resident involvement on mid-urethral sling procedures at our institution captured the resident-specific effect due to the new implementation of the residency training program in 2011 [10]. Our data with U-RAS demonstrated consistent findings, thereby adding support to the existence of a July Effect for operative time in urologic surgery.

Despite our findings, literature from other surgical specialties (plastic surgery and lower-extremity orthopedic trauma) [11, 22] did not identify any significant seasonal differences in operative time. Our findings may reflect a more difficult learning curve for laparoscopically naïve residents and interns, which could translate to longer operative times during their initial learning curve. A study in 2003 investigating subjective incident reports across three teaching hospitals found that the most commonly cited factor for surgical errors was inexperience/lack of competence on a surgical task (53% of incidents), followed by communication breakdowns (43%) and fatigue or excessive workload (33%) [23]. As the newly incoming resident surgeon is expected to participate in relatively high-level and unfamiliar operations, more mistakes likely occur which are intercepted by the attending surgeon prior to any event of tangible operative consequence (i.e., intra- or post-operative complications). For single-console robotic surgery, time is significantly impacted when the attending must switch places with the resident seated on the console to resolve a mistake or misstep by the resident. Furthermore, various systematic factors in robotic surgery, such as a lapse in teamwork or communication breakdowns that affect the surgical flow, could have a greater-than-usual influence on surgery times during this particular resident transition period. For example, flow disruptions (deviations in the natural progression of a surgical task) are increased in robotic cases with residents [24, 25].

Our data revealed that residents introduced significant variability to ST when we compared the different operative time intervals across cases stratified by complexity (Fig. 1a). ST remained stable in cases with attending surgeons only (prior to the implementation of our residency program in 2011). These findings might be partially explained by differences in the surgical technical skills of the incoming residents. Factors such as manual precision and dexterity, as well as responsiveness to operative instruction, are not readily weighed in the residency match process compared to metrics of academic performance (i.e., percentile class ranking, involvement with research) [26], which potentially leads to considerable differences in the technical skillset among the incoming residents. Additionally, the variation in ST significantly related to resident involvement suggests that the degree of resident impact on operative time may vary by case complexity. Radical cystectomies (Group 3) resulted in the greatest ST difference when comparing times with and without trainees, and they are the most technically challenging group of cases in this study. Increased complexity inherently comes with more attending-to-resident intraoperative education that contributes to the total operative time. However, the impact of resident trainee presence on high-complexity cases is likely to be more nuanced. A 2017 review found that the learning curve for robotic-assisted cystectomies was shorter compared to the learning curve for robotic-assisted laparoscopic prostatectomies (a lower complexity case) [27], which is likely due to the fact that trainees most often take on complex cases only after attaining competency in cases that are determined to be less technically challenging. Therefore, technical inexperience in high-complexity cases may not be a substantial contributing factor to delayed operative time.

Another notable finding is the reduced non-OR time (NT) due to resident involvement by an estimated 4.6 min (p < 0.001). The small, yet significant, shortened time could possibly reflect improved workflow when it comes to accomplishing surgical tasks that are more standard and consistently performed across surgical cases. This can be understood by looking more closely at the processes that occur during NT. Time from operating room entry to initial incision was shorter, possibly reflecting a trainee’s contribution to the surgical team with proper patient positioning, foley catheter insertion, and preparing the patient in the usual sterile fashion. Time also improved from completion of the procedure to exit from the operating room, a finding which could reflect contributions to the turnover process, such as transferring the patient. Our results demonstrate that the inefficiencies found in the operative time due to resident involvement may be partially offset with an improved non-operative time, but unanswered questions remain in understanding the factors that affect operative times when residents are involved.

While any combination of the above could account for the complex, likely multifactorial, explanation behind the findings of the current study, our finding of significant operative delay associated with residents in the earlier months of the academic year should prompt consideration of interventions and changes for RAS training. Recently, the use of virtual reality (VR) simulation has been gaining traction to attempt shortening the robotic surgery learning curve without compromising patient safety [28]. Randomized trials have found improved RAS skills and performance in the operating room with VR-simulated practice [29]. Formally integrating VR simulation into training may allow for junior residents to practice surgical skills in a more controlled environment. Developing a means to objectively evaluate RAS trainees in a simulated setting is a key step to validating such curricula and incorporating them into surgical training programs.

Limitations

The associations between operative variables and outcomes presented here do not necessarily prove causality. The variable-outcome association can be confounded by patient comorbidities and the operative experience of the resident surgeon. The multivariable models used in this study factored in patient age, BMI, case complexity, and attending surgeon experience; however, we were not able to adjust for resident experience, largely due to a limitation of the database concerning large inconsistencies with recording post-graduate year (PGY) level. Another limitation lies with using appropriate measure(s) for surgical outcomes, which often vary by the operation performed [30]. Besides EBL, post-operative length of stay and complications have been used in some studies and should be analyzed as an extension of this study. Lastly, our study was limited to a single center, but we suspect that there is broad generalizability to other institutions with similar practices.

Conclusion

We found that resident involvement in U-RAS is associated with prolonged operative time, with the greatest delay occurring in cases performed from July to October. There may be a July Effect on operative and healthcare efficiency for patients undergoing urologic RAS early in the academic year. Cases with residents after October (cases from November to June) demonstrated significant improvement in the overall surgical duration, suggesting that the laparoscopically naïve resident surgeons become well-acclimated to robotic surgery and the overall surgical flow. Proactively addressing the training needs of the incoming residents might mitigate their impact on surgical times; further investigations are needed to evaluate any targeted interventions.