Transfusion Risk in Open, Laparoscopic, and Robotic-Assisted Surgery: A Propensity Score Matched Case-Control Study across Surgical Disciplines

Patient blood management (PBM) is a critical aspect of modern surgical care, aiming to optimize patient outcomes by addressing hemostasis, blood conservation, and transfusion practices throughout the perioperative period. This multifaceted approach acknowledges the intricate balance between maintaining adequate blood volume and minimizing unnecessary blood loss during surgical procedures. Effective perioperative PBM strategies not only contribute to improved patient safety but also enhance recovery outcomes by mitigating the risks associated with anemia, transfusion-related complications, and prolonged hospital stays [1]. As advancements in surgical techniques continue to evolve, the integration of minimally invasive surgery has emerged as a promising avenue for optimizing perioperative blood management. It entails the execution of procedures via one or more small incisions, employing small tubes, miniature cameras, and surgical instruments. Numerous studies have documented improved outcomes in minimally invasive surgery over open surgery. Benefits have been well-documented, particularly in terms of decreased pain, shortened hospitalization periods, and reduced occurrence of complications [2‒11]. Increasingly surgical procedures are further refined with robotic surgical systems. Robotic-assisted surgery involves the use of robotic systems to aid surgeons in performing minimally invasive procedures. These systems typically consist of robotic arms controlled by a console where the surgeon sits and controls the instruments. The robotic arms are equipped with surgical tools and the surgeon operates them with precision using a console that provides a 3D view of the surgical site. Robotic-assisted surgery is considered to further improve patients outcomes due to its magnified vision and high precision [12]. By utilizing smaller incisions and precise instrumentation, robotic-assisted surgery could further reduce tissue trauma, thereby aligning with the principles of effective perioperative PBM. However, whether this leads to reduced perioperative blood loss and subsequently decreased blood transfusion needs across surgical disciplines remains uncertain.

While robotic-assisted surgery is commonly linked to enhanced outcomes compared to open procedures [12, 13], documenting the advantages of robotic-assisted surgery has proven more challenging when contrasted with laparoscopic procedures. Robotic surgery was first established for prostatectomies and has become the favored surgical procedure for oncological removal of the prostate since. While radical prostatectomy is the best studied procedure for robotic-assisted surgery so far, only few studies have demonstrated advantages of robotic-assisted surgery over laparoscopic surgery [14]. For other procedures, the evidence is even more limited. Especially for outcomes related to PBM, there is limited evidence to compare robotic-assisted surgery, laparoscopic surgery, and open surgery.

Studies comparing robotic surgery, laparoscopic surgery, and open surgery have shown advantages in terms of blood loss in only a few procedures, including robotic-assisted pancreatectomy, hepatectomy, and mesorectal excision [15‒17]. Whether the decreased intraoperative blood loss translates to a reduced perioperative requirement for allogeneic blood transfusions and improved postoperative hemoglobin levels remains uncertain. Generally, studies examining the effect of surgical modality on perioperative hemoglobin and transfusion requirements are lacking. The aim of this study was to compare the effect of robotic-assisted surgery, laparoscopic surgery and open surgery on allogeneic blood transfusion requirements, perioperative hemoglobin change and length of hospital stay.

This case-control study included patients aged ≥18 years who underwent robotic-assisted surgery as well as matched laparoscopic surgeries and open surgeries that were discharged from 14 German hospitals between January 1st, 2010, and December 31st, 2019. Patients without measures of preoperative and postoperative hemoglobin levels were excluded from the study. Surgeries with robotic-assisted systems that were not performed at least routinely within the German PBM Network hospitals, were not considered for this study. Detailed methods on the German PBM Network have been published previously [18]. Briefly, anonymized data were individually gathered from the electronic systems of participating hospitals, and supplemented with pharmacy and blood bank data by the local information technology officers. Routine error checks and validation were performed by center-specific experts and PBM Network biostatisticians. The analyzed outcomes were red blood cell (RBC) transfusions, perioperative hemoglobin change, and length of hospital stay.

Based on the procedure records, surgeries were classified as either robotic-assisted surgery, laparoscopic surgery or open surgery. Procedure details were documented in the form of electronic operation and procedure codes (OPS). OPS codes are very reliable for the documentation of medical procedures in Germany since they are nationally managed for billing, documentation, and statistics. Robotic surgery was defined as procedures with the OPS code 5-987. The included surgical procedures are listed in Table 1. Laparoscopic surgeries are defined in online supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000540981). Conversion to open surgery was classified as open surgery. Patients with robotic-assisted surgery were separately matched to patients with open surgery and patients with laparoscopic surgery to minimize confounding and ensure that patients undergoing different surgical procedures were comparable. Strictly one-to-one matching was performed in the absence of balanced groups. Nonparametric matching methods from the MatchIt package [19] in R were used. Pairwise matching was performed on age, sex, the Charlson comorbidity index and preoperative hemoglobin levels. Exact matching was performed for the type of surgery. This implies that, for example, patients with robotic-assisted radical prostatectomy were matched only with open or laparoscopic radical prostatectomy. Regarding the other confounders, matching was performed with optimal pair matching. As a distance measure between samples, propensity scores were estimated with logistic regression. An optimal match was defined by minimizing the sum of the absolute pairwise distance in the matched samples. Imbalances of matched samples were measured by the standardized mean differences and the average absolute within-pair difference statistics [20]. Values of standardized mean difference and pair difference close to zero indicate good balance in the matched sample. Alternative statistical approaches (e.g., weighting) were explored but yielded similar results. For ease of interpretability, this study presents only results from pairwise propensity score matching. Confounders were selected by acyclic graphing based on literature and expert consensus. Further analysis was performed on matched patients with logistic regression for RBC transfusion and linear regression for perioperative hemoglobin and length of hospital stay. The model estimates were further adjusted for time of discharge and hospital to minimize any effects over time and differences between medical centers. Graphing and statistical analysis were conducted using R software version 4.3.2. This study was reported following the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Out of 1,201,817 patients [18], 39,946 patients were eligible for the study (n = 5,843 robotic-assisted surgeries, n = 8,268 laparoscopic surgeries, and n = 25,835 open surgeries, Fig. 1). Characteristics of the patient cohort are shown in Table 2. Most frequent were surgeries on the large intestine (27.5%), kidney (19.3%), lung (17.6%), prostate (14.9%), and small intestine (12.0%). 15,009 matched patient pairs from propensity score matching were further included in the analysis. This population of matched patients was highly comparable regarding confounders. While the preoperative hemoglobin was very similar between matched pairs, postoperative hemoglobin varied distinctly between robotic-assisted surgery, laparoscopic surgery, and open surgery (Fig. 2). Transfusion rates per 1,000 patients were highest in open surgery on the large intestine (2,569 ± 304 RBC units), the small intestine (2,152 ± 750 RBC units), and lung (1,642 ± 274 RBC units). The lowest transfusion rates per 1,000 patients were consistently observed for radical prostatectomy with 300 (±53) for open surgery, 179 (±51) for robotic surgery and 147 (±149) for laparoscopic surgery (Fig. 3). While the RBC transfusion rate in robotic-assisted surgery on the small intestine was rather high, the overall use of RBC transfusions in this group was very low because of the limited number of patients (Fig. 4). Overall plasma and platelet transfusion patterns followed RBC transfusion patterns.

Comparing robotic versus open surgery 4,380 pairs of matched patients were analyzed (Fig. 1; Table 3), while 1,463 patients with robotic-assisted surgery and 21,455 patients with open surgery were left unmatched. RBC transfusion in robotic-assisted surgery was associated with a relative risk of 0.32 (95% CI: 0.27–0.37) and lead to significantly less reduction in perioperative hemoglobin levels in comparison to open surgery. Postoperative hemoglobin was 0.40 g/dL (95% CI: 0.31–0.49) higher in robotic-assisted surgery. The length of hospital stay was reduced by 4.29 days (95% CI: 3.74–4.84).

Comparing robotic versus laparoscopic surgery, 2,361 matched pairs were analyzed, while 3,482 patients with robotic-assisted surgery and 5,907 patients with laparoscopic surgery were left unmatched (Fig. 1; Table 3). Robotic-assisted surgery did not significantly reduce the risk of RBC transfusion compared to laparoscopic surgery (RR: 0.94 95% CI: 0.75–1.18). Postoperative hemoglobin was 0.27 g/dL (95% CI: 0.16–0.38) lower in robotic-assisted surgery. The length of hospital stay was not different: 0.53 days (95% CI: −0.14–1.19).

Comparing laparoscopic surgery versus open surgery, 8,268 matched pairs were analyzed, while 0 patients with laparoscopic surgery and 17,567 patients with open surgery were left unmatched (Fig. 1; Table 3). RBC transfusion in laparoscopic surgery was associated with a relative risk of 0.37 (95% CI: 0.34–0.40) and lead to significantly less reduction in perioperative hemoglobin levels compared to open surgery. Postoperative hemoglobin was 0.84 g/dL (95% CI: 0.79–0.90) higher in laparoscopic surgery. The length of hospital stay was reduced by 6.28 days (95% CI: 5.78–6.79) compared to open surgery.

In this large database analysis of the German PBM network including 42,566 patients, this case-control study showed that open surgery was associated with increased RBC transfusion requirements and higher perioperative hemoglobin reductions compared to both laparoscopic and robotic-assisted surgery. Between robotic-assisted and laparoscopic surgeries only for perioperative hemoglobin a small difference favoring laparoscopic surgery was observed. This study therefore supports the advantage of both robotic surgery and laparoscopic surgery over open surgery for improving perioperative PBM.

Most studies evaluate robotic surgery with a focus on estimated blood loss rather than RBC transfusions and perioperative hemoglobin. Altogether there is limited evidence to assess the impact of robotic surgery on outcomes relevant for PBM. Therefore, we evaluated perioperative hemoglobin to better quantify perioperative blood loss and RBC transfusion requirements to assess relevant perioperative anemia. The risk of allogeneic blood transfusions might be of special interest in cancer surgery since a detrimental effect on the recurrence of curable cancers has been suggested [21].

Consistently most studies on robotic-assisted surgery have shown a longer operative time compared to both open surgery and laparoscopic surgery [15]. Robotic-assisted surgery was also associated with higher treatment costs [22]. However, robotic-assisted surgery significantly reduced the length of hospital stay compared to open surgery and could further decrease treatment costs by reducing perioperative complications rates. To quantify the extent of this reduction in treatment costs due to decreased perioperative complication rates remains challenging. Of note, outcomes in robotic-assisted surgery depend on the specific procedure, surgeon experience, patient characteristics, and other factors. Additionally, advancements in robotic technology and surgical techniques may influence these findings over time.

Robotic-assisted radical prostatectomy is one of the most common applications of robotic surgery. Research indicates that robotic-assisted radical prostatectomy may be associated with reduced blood loss and shorter hospital stays compared to open surgery [23, 24]. However, in the first trial comparing laparoscopic radical prostatectomy versus robotic-assisted radical prostatectomy, laparoscopic radical prostatectomy was associated with lower blood loss [14]. In our study, radical prostatectomy was also the only evaluated procedure with more robotic surgeries performed than laparoscopic or open surgeries. This further highlights that robotic surgery has become the standard of care for radical prostatectomy in the evaluated population. In radical prostatectomy, the a priori risk of allogeneic blood transfusions is very low [14, 23]. In other fields of surgery with higher risk of allogeneic blood transfusion, a more distinct effect of robotic surgery can be expected. In addition to prostate surgery, robotic-assisted procedures for other urological conditions have been examined. Studies have reported variable results in terms of blood loss, with some indicating advantages for robotic surgery. For both radical and partial nephrectomy, studies have repeatedly found no difference in blood loss and transfusion requirements between robotic-assisted and laparoscopic approaches [25‒27]. This is consistent with the findings in our study.

Studies in colorectal robotic surgery have shown inconclusive results regarding blood loss and transfusion requirements. Some research articles suggest comparable blood loss between robotic and laparoscopic procedures, while others report lower blood loss in robotic surgery [15, 22, 28]. One study that showed lower blood loss for robotic-assisted surgery also reported on postoperative hemorrhage and blood transfusion requirements with no significant difference between robotic, laparoscopic, and open surgery [15]. While our study was unable to report details on blood loss, we found that both, robotic-assisted surgery and laparoscopic surgery versus open surgery were associated with a decreased risk of RBC transfusion and limited perioperative hemoglobin reduction.

In thoracic surgery, decreased bleeding has been suggested for laparoscopic surgery versus open surgery [11]. Similar blood loss has been suggested between robotic-assisted and laparoscopic surgery [29]. However, fewer studies have evaluated blood loss in this field of surgery [30, 31]. Even fewer studies have investigated blood loss in gastric bypass surgery with no evidence for reduced blood loss in robotic-assisted surgery [32]. In both fields of surgery, this study provides to our knowledge first evidence for a reduced risk of allogeneic blood transfusion and limited perioperative hemoglobin reduction.

Robotic-assisted surgery could be advantageous in other fields of surgery that were not included in this study due to lack of routine performance in the included study centers. Procedures of interest could be surgery on the pancreas, liver, and esophagus. Robotic-assisted or laparoscopic pancreatectomy and hepatectomy has been suggested to reduce blood loss [16, 17]. In a single-center randomized controlled trial robot-assisted, minimally invasive thoracolaparoscopic esophagectomy has been shown to reduce blood loss compared to open transthoracic esophagectomy, while comparisons to laparoscopic approaches are lacking [33].

Limitations of our study are mainly driven by its retrospective nature. While we aimed to minimize confounding and maximize comparability of the three groups with propensity score matching, the potential for unmeasured confounders remains. This study did not account for surgeon experience. Studies have shown that outcomes improve with surgical skill training, specialization, and years of experience [34]. However, other studies suggest that robotic-assisted surgery may eliminate physician- and hospital-related factors [35]. The statistical approach of propensity score matching ensured high internal validity of our findings but may limit generalizability due to the exclusion of patients. The number of included patients varied by field of surgery. This variation was mostly due to a lack of adoption of minimally invasive surgical techniques. Ultimately, this approach led to a higher number of unmatched patients. While results including unmatched patients were similar, the results from propensity score matching are deemed to be more reliable due to the diversity of included surgical procedures. Studies have suggested that the culture within specific specialties and the target anatomic areas are major influences for the adoption of robotic-assisted surgery [36]. Other limitations include the absence of data regarding post-discharge follow-up. Thus, we were unable to report on other endpoints of interest including post-discharge hemoglobin changes. Strengths of this study include the large sample size and multicenter nature of the German Patient Blood Management network.

In conclusion, our study underlines the benefits of both robotic-assisted surgery and laparoscopic surgery to reduce blood loss and the need for RBC transfusions when compared to open surgery. This highlights the potential of minimally invasive procedures to significantly enhance patient outcomes. However, between robotic-assisted surgery and laparoscopic surgery no clear-cut benefit for either of the two was observed. Overall, the integration of robotic technology and minimally invasive techniques represents a significant advancement in surgical practice, offers improved outcomes, and minimizes the inherent risks associated with invasive procedures including transfusion of blood products.

We would like to thank the information technology officers of every participating hospital for their support in this project by providing the necessary data.

Approval was obtained by the leading Ethics Committee of the University Hospital Frankfurt (Reference 318/17), by the Ethics Committees of all participating centers, and by the Hessian data protection officer (Reference 43.60; 60.01.21-ga; October 24, 2018). Any written informed consent of patients was waived by the Ethics Committee.

P.M. or his department received research grants from the German Research Foundation (ME 3559/1-1, ME 3559/3-1, and ME 6094/3-2), Bundesministerium für Bildung und Forschung (01KG1815), and Bundesministerium für Gesundheit (ZMVI1-2520DAT10E), and honoraria for scientific lectures from Abbott, Aesculap Academy, B. Braun Melsungen, Biotest AG, Vifor Pharma, Ferring, CSL Behring, German Red Cross/Institute of Transfusion Medicine, HCCM Consulting, Löwenstein Medical, HemoSonics, Pharmacosmos, and Siemens Healthcare. K.Z. and his department received unrestricted grants from B. Braun Melsungen, Fresenius Kabi, CSL Behring, and Vifor Pharma for the implantation of PBM in the University Hospital Frankfurt; awards from Aktionsbündnis Patientensicherheit, European Society of Anaesthesiology and Intensive Care, Lohfert Stiftung, Masimo Patient Safety Foundation, and MSD Gesundheitspreis; and honoraria or travel support for consulting or lecturing from Abbott, Aesculap Akademie, AQai, Astellas Pharma, AstraZeneca, Aventis Pharma, B. Braun Melsungen, Baxter Deutschland, Biosyn, Biotest, Bristol Myers Squibb, CSL Behring, Dr. Franz Köhler Chemie, Dräger Medical, Essex Pharma, Fresenius Kabi, Fresenius Medical Care, Gambro Hospal, Gilead, GlaxoSmithKline, Grünenthal, Hamilton Medical, HCCM Consulting, Löwenstein Medical, Janssen-Cilag, medupdate, Medivance Europe, MSD Sharp & Dohme, Novartis Pharma, Novo Nordisk Pharma, P. J. Dahlhausen & Co., Pfizer Pharma, Pulsion Medical Systems, Siemens Healthcare, Teleflex Medical, Teva, TapMed Medizintechnik, Verathon Medical, and Vifor Pharma. The results in this publication are thematically linked but not directly related to specific activities of the projects ENVISION and COVend, which have received funding from the European Union’s Horizon 2020 and Horizon Europe research and innovation programmes, respectively, under grant agreement No. 101015930 (ENVISION) and No. 101045956 (COVend). AUS is part of the German Research Foundation (Deutsche Forschungsgemeinschaft), research group FerrOs (STE 1895/9-1 and STE 1895/10-1) and receives a research grant from Pharmacosmos to perform a single-center prospective trial on preoperative anemia treatment. J.T. received honoraria for scientific lectures from Vifor Pharma Germany. P.F. received honoraria for scientific lectures from Medtronic, and honoraria for scientific lectures and consultancy from Dräger. M.G. received honoraria or travel support for consulting or lecturing from CSL Behring, Edwards Lifesciences, Fresenius Medical Care, G.E. Healthcare, Grünenthal, Johnson & Johnson, Medtronic and Vifor Pharma. All other authors declare no conflict of interest.

The authors did not receive any funding for the preparation of the data or the manuscript.

Study conception/design and clinical interpretation: P.M., F.R., and K.Z. Data acquisition: A.B., A.R., A.U.S., D.N., G.E., H.W., J.F., J.T., M.G., M.V., O.B., and P.F. Data analysis: F.R. and P.M. Statistical interpretation and writing of draft manuscript: F.R., P.M., S.C., L.H., and K.Z. Revision/approval of the final manuscript: all authors.

German Patient Blood Management Network Collaborators: See supplementary table 2.

For original data, contact [email protected].