Importance
As robotic-assisted cardiac surgical procedures increase nationwide, surgeons need to be educated on the safety of the new modality compared with that of open technique.
Objective
To compare complications, length of stay (LOS), actual cost, and mortality between nonrobotic and robotic-assisted cardiac surgical procedures.
Design, Setting, and Participants
Weighted data on cardiac patients who had undergone operations involving the valves or septa and vessels, as well as other heart and pericardium procedures, from January 1, 2008, to December 31, 2011, were obtained from the Nationwide Inpatient Sample via the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality. Propensity score matching was used to match each robotic-assisted case to 2 nonrobotic cases on 14 characteristics.
Main Outcomes and Measures
Complications, median LOS, actual cost, and mortality.
Results
Exploratory analysis found a total of 1 374 653 cardiac cases (1 369 454 [99.6%] nonrobotic and 5199 [0.4%] robotic-assisted cases). After propensity score matching, there were 10 331 (66.5%) nonrobotic cases and 5199 (33.5%) robotic-assisted cases. Cardiac operations included 1630 (10.5%) involving the valves or septa, 6616 (42.6%) involving the vessels, and 7284 (46.9%) other heart and pericardium procedures. Robotic-assisted compared with nonrobotic surgery had a higher median cost ($39 030 vs $36 340; P < .001) but lower LOS (5 vs 6 days; P < .001) and lower mortality (1.0% vs 1.9%; P < .001). Robotic-assisted surgery had significantly fewer complications for all operation types (30.3% vs 27.2%; P < .001).
Conclusions and Relevance
Overall, robotic-assisted surgery has significantly reduced median LOS, complications, and mortality compared with nonrobotic surgery. Results of this study support the contention that robotic-assisted surgery is as safe as nonrobotic surgery and offers the surgeon an additional technique for performing cardiac surgery.
Similar to laparoscopy, with its advent in the 1980s, progression, and acceptance through the 1990s, robotic-assisted surgery has followed a similar rugged path. Since its conception, the symbiotic relationship between robot and surgeon, allowing smooth and minute movements and 3-dimensional vision, has added greatly to the field of surgery.1 Robotic-assisted surgery attempts to improve on laparoscopic surgery by providing increased intracavity articulation, increased degrees of freedom, and downscaling of motion amplitude, which may reduce the strain on the surgeon.2,3 For these reasons, in general surgery and subspecialties, the use of the robot has increased significantly during the past 5 years.
The biggest growth in robotic-assisted surgery has been seen in the fields of gynecology and urology. Recently, Wright et al4 reported an increase in robotic-assisted hysterectomy from 0.5% of the procedures in 2007 compared with 9.5% in 2010 for benign disease. In their study, robotic-assisted surgery had similar outcomes to laparoscopic surgery; however, robotic-assisted surgery had a higher total cost of $2189 more per case. Similarly, in urologic surgery, Leddy et al5 reported in 2010 that radical prostatectomy remains the biggest use of robotic-assisted surgery in urology, with 1% in 2001 to 40% of all cases in 2006 performed in the United States.
Specific to cardiac surgery, as early as 1999, the advantages of the robot in coronary artery bypass grafting and valvular operations were demonstrated with increased visualization, ease of harvest, and quality of vascular anastomoses.6-8 Although safety and efficacy of robotic-assisted surgery are supported, given the high cost of the robot itself, longer operating times, and short life of the robotic instruments, it remains to be established whether the robot is cost-effective.9-11 The purpose of the present study was to compare outcomes of complications, length of stay (LOS), actual cost, and mortality between nonrobotic and robotic-assisted cardiac surgical procedures.
Discharge data from January 1, 2008, to December 31, 2011, were obtained from the Nationwide Inpatient Sample (NIS) via the Healthcare Cost and Utilization Project (HCUP) of the Agency for Healthcare Research and Quality. The NIS consists of a 20% stratified sampling of US community hospital discharges and is the largest source of hospital discharge information in the United States. It compiles various demographics, including admission and discharge data, hospital data, total charges, actual cost, and other characteristics of discharges (http://www.hcup-us.ahrq.gov/nisoverview.jsp).
The present study used weighted HCUP data to provide approximate total national statistics. Weighted data also help assure that national estimates of hospitalizations and hospitalization rates are comparable across years despite the varying number of states participating in each year of the HCUP. In addition, using the HCUP-NIS Cost-to-Charge Ratio Files, charges were able to be converted to cost. Charges are the amount that each hospital bills for services, whereas cost is the amount that the services actually cost to provide (http://www.hcup-us.ahrq.gov/db/state/costtocharge.jsp#overview). Using “cost-to-charge ratios allows the translation of total charges into actual costs using a validated conversion factor that provides an estimate of all-payer inpatient costs. Wide variation in hospital prices between different regions makes cost-to-charge adjustments in reported hospital costs necessary.”12(p2320) To this end, actual costs are reported in this article. We completed a data use agreement with HCUP-NIS, and the study was considered exempt by WellSpan Health’s Institutional Review Board.
All study variables were initially defined and coded by the NIS. Weighted HCUP-NIS data were from January 1, 2008, to December 31, 2011. Cardiac surgical procedures were identified using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), procedure codes; only cardiac cases with ICD-9-CM codes in procedures 1 through 15 were included. Cardiac operations included valves or septa (codes 35.1-35.95), vessels (36.03-36.99), and other heart and pericardium procedures (37.11-37.74). Aortic (3511, 3521, 3522, 3539, 3541, 3551, 3561, and 3571) and mitral (3512, 3523, and 3524) procedures were used for subset analysis. Robotic-assisted cases were identified by using ICD-9-CM procedure codes 17.41, 17.42, 17.43 and 17.44; cases were identified as robotic-assisted or nonrobotic procedures as appropriate.
Similar to Brunt et al,13 the present study propensity matched robotic-assisted to nonrobotic procedures according to patient characteristics, comorbidities, and hospital characteristics. Regarding comorbidities, the Charlson Comorbidity Index score was the propensity-matched variable; this score takes into account conditions such as myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic obstructive pulmonary disease, connective tissue disease, peptic ulcer disease, diabetes mellitus, moderate to severe chronic kidney disease, hemiplegia, leukemia, malignant lymphoma, solid tumor liver disease, and AIDS. For the present study, Charlson Comorbidity Index scores were calculated using ICD-9-CM diagnosis codes according to the Charlson et al14 and Deyo et al15 validated and published methods.
Propensity score matching was used to compare each robotic-assisted case with 2 nonrobotic cases (ie, a 1:2 ratio to help control for variances resulting from size differences between the 2 groups) on 14 characteristics (age, sex, race, payer, elective vs nonelective surgery, Charlson Comorbidity Index score, hospital bed size, location, region and teaching status, annual income, and the 3 operation subtypes). The matching variables for propensity matching were chosen based on the significant differences between nonrobotic and robotic-assisted surgery (discussed in the Results section). Outcome variables were median LOS, actual cost, and mortality. Complication variables were aggregated into “complication yes or no”16 (yes indicating having 1 or more of any of the complications) and total number of complications (sum of all the complications per patient). Nonparametric analyses were conducted because most outcome variables were not normally distributed; therefore, median complications, LOS, and in-hospital cost were reported. All variables used in the present study, including complications and mortality, are associated with the index admission.
Exploratory analyses were conducted using descriptive statistics (frequencies) and χ2 or Fisher exact tests to explore associations between robotic-assisted or nonrobotic surgery and patients’ characteristics. Comparative analysis of outcomes used nonparametric statistics such as Kruskal-Wallis and Mann-Whitney tests to explore differences between robotic-assisted surgery and outcomes.
Two multiple regressions were performed to predict cost and mortality using type of surgery (nonrobotic or robotic-assisted), sex (male or female), age (<18, 18-69, 70-79, and ≥80 years), race (white or nonwhite), payer (Medicare, Medicaid, private, or other), elective or nonelective surgery, Charlson Comorbidity Index score, hospital size (small, medium, or large), hospital location (urban or rural), hospital region (Northeast, Midwest, South, or West), teaching status of hospital (teaching or nonteaching), and operation type (valve or septa, vessel, other) as our predictors. P ≤ .05 was considered statistically significant; all P values were reported as 2-sided. Because the NIS is a 20% stratified sampling, all analyses were weighted by discharge. All statistical analyses were performed using SPSS statistical software, version 20 (IBM Corp). Propensity matching was performed using Propensity Score Matching for SPSS, version 3.0 (http://arxiv.org/ftp/arxiv/papers/1201/1201.6385.pdf).
Exploratory analysis found a total of 1 374 653 cardiac cases comprising 1 369 454 (99.6%) nonrobotic and 5199 (0.4%) robotic-assisted cases. Use of robotic-assisted surgery increased during the study period, from 0.057% in 2008 to 0.390% in 2011 (P < .001). Table 1 summarizes patient demographics, comorbidities, and hospital characteristics before propensity score matching. Before propensity matching, there was a disparity between demographics, comorbidities, hospital characteristics, and timing of operation (elective vs nonelective) between patients undergoing nonrobotic and robotic-assisted surgical procedures, which could influence outcomes. Robotic-assisted surgery was associated with male sex (3582 [68.9%]), white race (3347 of 3997 [83.7%]), private insurance (2632 [50.6%]), elective surgery (3952 of 5189 [76.2%]), large hospitals (4298 of 5180 [83.0%]), hospitals in urban locations (5145 of 5180 [99.3%]), hospitals in the Midwest (1749 [33.6%]) and West (1162 [22.4%]), teaching hospitals (4392 of 5180 [84.8%]), and median household income greater than $63 000 (1770 of 5119 [34.6%]).
Table 2 summarizes patient demographics, comorbidities, and hospital characteristics after propensity score matching, showing that there are no significant differences between characteristics of patients undergoing nonrobotic and robotic-assisted surgery, which reduces confounders that could influence outcomes. After propensity matching, there were 10 331 (66.5%) nonrobotic cases and 5199 (33.5%) robotic-assisted cases. Overall, cardiac operations included 1630 (10.5%) involving the valves or septa, 6616 (42.6%) involving the vessels, and 7284 (46.9%) other heart and pericardium procedures.
Comparative univariate analysis revealed that robotic-assisted surgery, compared with nonrobotic surgery, had higher cost ($39 030 vs $36 340; P < .001), but lower LOS (5 vs 6 days; P < .001) and mortality (1.0% vs 1.9%; P < .001). For those who had 1 or more complication, robotic-assisted surgery had fewer complications (1414/5199 [27.2%]) compared with nonrobotic surgery (3129/10331 [30.3%]; P < .001) (Table 3). This outcome was seen regardless of the operation type.
Subset analysis indicated that robotic-assisted aortic valve, mitral valve, and cardiac vessel surgery was more costly than nonrobotic surgery, but had lower mortality and LOS than nonrobotic surgery for all surgery types (Table 4). Robotic-assisted elective and nonelective surgery cost more than their nonrobotic counterparts (elective, $38 513 vs $34 602, P < .001; nonelective, $42 353 vs $41 185, P = .002).
Multiple regression used to assess the association between several independent variables to predict cost (eTable in the Supplement) found that having nonrobotic surgery decreased the cost by $1531. Regardless of the type of surgery, nonelective surgery added $8397. This analysis takes into account the interaction of covariates and their combined effect on cost.
Multiple regression for mortality indicated that robotic-assisted surgery and elective surgery were not predictors of mortality (robotic-assisted surgery, odds ratio, 0.627; elective surgery, odds ratio, 0.580). Predictors of higher risk of mortality were rural location of hospital, age 80 years or older, Charlson Comorbidity Index score of 4 or more, hospital located in the Midwest, and teaching hospital (Table 5).
It is well understood that advancement of technology drives the growth of cost in health care. It is often argued that patients and physicians adopt new modalities of treatment before their weaknesses or advantages are clear. Between 2007 and 2009, there was a 75% increase in the number of DaVinci systems purchased in the United States.10 The number of DaVinci systems purchased outside the United States had doubled in that period. Debate exists as to whether use of this technology is cost effective. Our study indicates a similar finding: during a 4-year study, robotic-assisted cardiac surgery increased 600%, from 0.06% to 0.4% for all cardiac operations.
Our study has shown significantly lower mortality on propensity matching in patients undergoing robotic-assisted surgical procedures. This outcome has been seen previously in benign gynecologic and urologic procedures. The lower mortality was seen in all 3 different types of cardiac operations. Our finding for mortality in the group undergoing robotic-assisted mitral valve surgery (1.2%) was similar to findings by Seco et al,17 with mortality ranging from 0% to 3.0%. Our study attributed the lower mortality to improved visualization by the surgeon with the robotic-assisted technology.
The lower mortality seen with robotic-assisted surgery in the study by Seco et al17 could be because it was more of an elective process, as also seen in our study. This scenario permits surgeons to be more careful in selecting their population to undergo robotic-assisted surgical procedures and allows patients’ comorbidities to be managed by their primary care physician prior to surgery compared with open operations performed in an emergency or urgent setting.
Although robotic-assisted surgery has become popular, the major controversy remains its cost. The controversy regarding the cost of robotic-assisted surgery is its high fixed cost of $1 million to $2.5 million (US $) per unit, the caseload of 150 to 250 procedures needed for surgeons to become versed in the use of the robot and the complications during this period,18 the fixed usage life of 10 procedures per robotic instrument, and the use of single-use consumables; in addition, the increased operating room time also robs the system of overall efficiency.10
Using multiple regression, the present study found that nonrobotic surgery decreased the cost by $1531, which is similar to results of prior studies that had cited the high fixed cost of robotic-assisted surgery. However, the higher cost of robotic-assisted surgery may be offset by the decrease in complications and LOS with overall improved efficiency.
Our study has shown that robotic-assisted surgery has lower overall complications and LOS. This advantage could lower overall cost, as seen in our multivariable analysis. Similarly, Jonsdottir et al19 have shown that minimally invasive operations, including robotic-assisted procedures, resulted in a significant decrease in complications without an increase in total mean cost. In urologic surgery, it was shown that costs with minimally invasive surgery, including robotic-assisted, can decrease with nephrectomy as efficiency increases, including shorter operating room times and LOS.20 This outcome was also seen in our study: LOS was significantly lower in patients undergoing robotic-assisted surgery (5 days) compared with those undergoing nonrobotic surgery (6 days).
The technological advances associated with robotic-assisted surgery, however, have increased patient quality of life, lowered pain scores, and in some cases, have decreased surgical complications.8,21 The cost of hospital stay, most notably time spent in the intensive care unit, is therefore also decreased secondary to shorter LOS. This, however, is a limitation in our study since our administrative database is unable to assess the shorter LOS in the intensive care unit.
Our study was also unable to assess the effects on cost of improved patient comfort from robotic-assisted surgery. One study performed at the University of Maryland showed that high-risk patients have the highest benefit with robotic-assisted cardiac valve operations, with longer graft patency and improved quality of life that may contribute to decreasing its cost.22 Therefore, several studies in the literature affirm that use of robotic-assisted surgery offset the high initial cost.
There are some limitations to our study. For instance, the surgeons’ fellowship training and number of procedures performed in training are unknown. However, the study showing the effect of hospital and surgeon volume was criticized for its methods and the quality of statistical analysis.23 In addition, our study is a retrospective observational study and carries potential selection biases associated with this type of study. It also lacks 30-day follow-up data. It does not address long-term outcomes, and longitudinal analysis in the form of randomized clinical trials is needed to determine favorable in-hospital outcomes for robotic-assisted cardiac surgery. Our database is primarily an administrative data set and relies mostly on nonclinical personnel to code patients, health care professionals, and procedural factors using ICD-9-CM classification. This setting may lead to coding errors, but propensity matching minimizes the effect of such errors and is helpful for controlling data. However, it does not control for surgeon bias, and we do not have long-term data, or prehospital data.24 Furthermore, when comparing ICD-9-CM codes between the adult cardiac surgery databases of the Society of Cardiothoracic Surgeons of Great Britain and Ireland and the Society of Thoracic Surgeons, Mack et al25 and Hickey et al26 were able to demonstrate that the latter offers more comprehensive accurate data on key cardiovascular and surgical variables such as technique, knowledge of anatomy, intraoperative complications, and operative times, which demonstrates a gap in data collection. Despite propensity matching in 14 characteristics, our data were limited by not including New York Heart Association class, ejection fraction, and renal dysfunction, as well as other comorbidities for reasons mentioned above. The lack of information as well as other relevant risk factors introduces an important systematic confounding variable that cannot be controlled for in this analysis.
However, while retrospective data have limitations, this study is a good way to look at safety before performing a prospective study and does not require a multi-institutional design.
Robotic-assisted surgery appears to have reduced mortality and complications,27 helping to offset upfront costs. Results of this study suggest that robotic-assisted surgery may be as safe as nonrobotic surgery and offer the surgeon an additional technique for performing cardiac surgery. Further studies need to be performed to show long-term benefits, including quality of life and pain scores. For 30-day follow-up, further research may also consider the use of the Society of Thoracic Surgeons database.
Accepted for Publication: March 11, 2015.
Corresponding Author: Vanita Ahuja, MD, MPH, Department of General Surgery, York Hospital, 1001 S George St, York, PA 17405 (vahuja@wellspan.org).
Published Online: June 17, 2015. doi:10.1001/jamasurg.2015.1098.
Author Contributions: Messrs Bell and Grim 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.
Study concept and design: Bell, Grim, Martin, Ahuja.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: Yanagawa, Grim, Martin, Ahuja.
Statistical analysis: Bell, Grim, Martin, Ahuja.
Administrative, technical, or material support: Yanagawa, Perez, Ahuja.
Conflict of Interest Disclosures: None reported.
1.Sackier
JM, Wang
Y. Robotically assisted laparoscopic surgery: from concept to development.
Surg Endosc. 1994;8(1):63-66.
PubMedGoogle ScholarCrossref 2.Cadière
GB, Himpens
J, Germay
O,
et al. Feasibility of robotic laparoscopic surgery: 146 cases.
World J Surg. 2001;25(11):1467-1477.
PubMedGoogle Scholar 4.Wright
JD, Ananth
CV, Lewin
SN,
et al. Robotically assisted vs laparoscopic hysterectomy among women with benign gynecologic disease.
JAMA. 2013;309(7):689-698.
PubMedGoogle ScholarCrossref 6.Loulmet
D, Carpentier
A, d’Attellis
N,
et al. Endoscopic coronary artery bypass grafting with the aid of robotic assisted instruments.
J Thorac Cardiovasc Surg. 1999;118(1):4-10.
PubMedGoogle ScholarCrossref 7.Felger
JE, Chitwood
WR
Jr, Nifong
LW, Holbert
D. Evolution of mitral valve surgery: toward a totally endoscopic approach.
Ann Thorac Surg. 2001;72(4):1203-1208.
PubMedGoogle ScholarCrossref 8.Falk
V, Diegeler
A, Walther
T,
et al. Total endoscopic computer enhanced coronary artery bypass grafting.
Eur J Cardiothorac Surg. 2000;17(1):38-45.
PubMedGoogle ScholarCrossref 9.Giulianotti
PC, Coratti
A, Angelini
M,
et al. Robotics in general surgery: personal experience in a large community hospital.
Arch Surg. 2003;138(7):777-784.
PubMedGoogle ScholarCrossref 10.Barbash
GI, Glied
SA. New technology and health care costs—the case of robot-assisted surgery.
N Engl J Med. 2010;363(8):701-704.
PubMedGoogle ScholarCrossref 11.Morgan
JA, Thornton
BA, Peacock
JC,
et al. Does robotic technology make minimally invasive cardiac surgery too expensive? a hospital cost analysis of robotic and conventional techniques.
J Card Surg. 2005;20(3):246-251.
PubMedGoogle ScholarCrossref 12.Lee
A, Johnson
JA, Fry
DE, Nakayama
DK. Characteristics of hospitals with lowest costs in management of pediatric appendicitis.
J Pediatr Surg. 2013;48(11):2320-2326.
PubMedGoogle ScholarCrossref 13.Brunt
ME, Egorova
NN, Moskowitz
AJ. Propensity score–matched analysis of open surgical and endovascular repair for type B aortic dissection.
Int J Vasc Med. 2011; 2011:364046. doi:
10.1155/2011/364046PubMedGoogle Scholar 14.Charlson
ME, Pompei
P, Ales
KL, MacKenzie
CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.
J Chronic Dis. 1987;40(5):373-383.
PubMedGoogle ScholarCrossref 15.Deyo
RA, Cherkin
DC, Ciol
MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases.
J Clin Epidemiol. 1992;45(6):613-619.
PubMedGoogle ScholarCrossref 16.LaPar
DJ, Bhamidipati
CM, Mery
CM,
et al. Primary payer status affects mortality for major surgical operations.
Ann Surg. 2010;252(3):544-550.
PubMedGoogle Scholar 17.Seco
M, Cao
C, Modi
P,
et al. Systematic review of robotic minimally invasive mitral valve surgery.
Ann Cardiothorac Surg. 2013;2(6):704-716.
PubMedGoogle Scholar 18.Krypson
AP, Nifong
LW, Chitwood
WR. Robot-assisted surgery: training and re-training surgeons.
Int J Med Robot.2004;1(1):70-76.
PubMedGoogle ScholarCrossref 19.Jonsdottir
GM, Jorgensen
S, Cohen
SL,
et al. Increasing minimally invasive hysterectomy: effect on cost and complications.
Obstet Gynecol. 2011;117(5):1142-1149.
PubMedGoogle ScholarCrossref 20.Alemozaffar
M, Chang
SL, Kacker
R, Sun
M, DeWolf
WC, Wagner
AA. Comparing costs of robotic, laparoscopic, and open partial nephrectomy.
J Endourol. 2013;27(5):560-565.
PubMedGoogle ScholarCrossref 21.Walther
T, Falk
V, Metz
S,
et al. Pain and quality of life after minimally invasive versus conventional cardiac surgery.
Ann Thorac Surg. 1999;67(6):1643-1647.
PubMedGoogle ScholarCrossref 25.Mack
MJ, Herbert
M, Prince
S, Dewey
TM, Magee
MJ, Edgerton
JR. Does reporting of coronary artery bypass grafting from administrative databases accurately reflect actual clinical outcomes?
J Thorac Cardiovasc Surg. 2005;129(6):1309-1317.
PubMedGoogle ScholarCrossref 26.Hickey
GL, Grant
SW, Cosgriff
R,
et al. Clinical registries: governance, management, analysis and applications.
Eur J Cardiothorac Surg. 2013;44(4):605-614.
PubMedGoogle ScholarCrossref 27.Wilt
TJ, MacDonald
R, Rutks
I, Shamliyan
TA, Taylor
BC, Kane
RL. Systematic review: comparative effectiveness and harms of treatments for clinically localized prostate cancer.
Ann Intern Med. 2008;148(6):435-448.
PubMedGoogle ScholarCrossref