In the current climate of increasing health care spending, many efforts aim to curb costs without compromising patient care. The Independent Payment Advisory Board, created under the Affordable Care Act in 2010, is one such effort; its sole purpose is to reduce costs through modification of Medicare reimbursement rates on the basis of clinical efficacy.1 In gynecology, the cost of various routes of hysterectomy will likely come under scrutiny because nearly 600,000 hysterectomies are performed annually in the United States.2
Approaches to hysterectomy include abdominal, vaginal, laparoscopic, and, more recently, robotic. As a primary route of hysterectomy, vaginal hysterectomy is the method recommended by the American College of Obstetricians and Gynecologists3,4 and has been shown to be less costly than either abdominal or traditional laparoscopic approaches.5–7 Advantages of the robotic approach have led health care practitioners to incorporate it into their practice, although several studies have shown that the robotic technique may be more expensive than traditional laparoscopy and laparotomy.8–12
Intuitively, vaginal hysterectomy should be less expensive than robotic hysterectomy. However, data on this topic are curiously lacking with only two studies directly comparing inpatient costs between the two methods.13,14 Because performance of robotically assisted hysterectomy is increasing in the setting of an evolving mandate to provide value-based health care, we aimed to evaluate the cost of robotic hysterectomy and compare it with abdominal and vaginal routes by using an all-cause cost model inclusive of all billed costs through 6 weeks after surgery. Our secondary objective was to estimate the changes in cost in the presence of operative complications.
MATERIALS AND METHODS
This study was approved by the Mayo Clinic institutional review board. We identified all robotically assisted hysterectomy cases, with or without bilateral salpingo-oophorectomy, that were performed at the Mayo Clinic (Rochester, Minnesota) since acquisition of the da Vinci system in gynecologic surgery from January 1, 2007, through December 31, 2009. We used Surgical Operative Note Explorer software, which searches all text in the operative note, regardless of adjacent letters or spaces, to identify those with the selected terms robotic and da Vinci. The identified operative notes were then manually reviewed to confirm robotic hysterectomy. We included any indication for hysterectomy, but cases of radical hysterectomy, hysterectomy with lymphadenectomy or cancer staging, and hysterectomy with any concomitant nongynecologic surgery other than appendectomy were excluded from the study. The resultant cohort of patients was then crossreferenced with a list generated by the institutional surgical information recording system to ensure accuracy and capture of all robotic hysterectomies.
For the comparison groups, the surgical information recording system was searched for cases of abdominal or vaginal hysterectomy with or without bilateral salpingo-oophorectomy. Only the 3 years immediately preceding acquisition of the da Vinci surgical system (2004–2006) were searched to decrease selection bias between groups, because robotically assisted hysterectomy was not an option during that time. A random sample of abdominal and vaginal hysterectomies was then generated from the identified cases. We followed the same exclusion criteria as for robotic cases but also excluded cases of vaginal hysterectomy with concurrent pelvic floor reconstructive surgery because this is a known risk factor for perioperative morbidity.15 All cases were performed by fellowship-trained gynecologic oncologists and urogynecologists.
Data on all surgical cases were then recorded using a standardized data collection form and abstracted from the medical record. Data collected included basic demographic information, body mass index, comorbid conditions, uterine weight, operative time, and intraoperative and postoperative complications until time of discharge. Operative time was defined as time from incision to closure and included robot docking. Intraoperative complications were defined as any injury to pelvic viscera or vessels; postoperative complications were defined as any deviation from the normal postoperative course.16 Binary indicators for comorbidities in major therapeutic areas were also abstracted.
Patients undergoing robotic hysterectomy were matched to patients undergoing abdominal or vaginal hysterectomy. Matches were established separately through one-to-one propensity score matching to ensure comparable baseline characteristics between patients undergoing robotic and abdominal hysterectomy and between patients undergoing robotic and vaginal hysterectomy. Propensity score is defined as “the conditional probability of being treated given the covariates.”15,17 Each robotic case was matched to a vaginal and abdominal hysterectomy case with a similar propensity score, which ensured that the observed covariates were similar among the cohorts. Because each group was matched by propensity score, the robotic cohorts were slightly different for each comparison group. Comparison of continuous covariates between the study cohorts was based on the rank sum test or t test, depending on whether the data were normally distributed, whereas categorical variables were compared using the χ2 test.
All-cause costs were captured from the Olmsted County Healthcare Expenditure and Utilization Database.18,19 Using claims data, the Olmsted County Healthcare Expenditure and Utilization Database collects information on resource use and the corresponding charges, which is standardized with inflation adjustment using 2009 constant dollars and also accounts for geographic wage differences. The Olmsted County Healthcare Expenditure and Utilization Database uses a “bottom-up” approach in which health care resource use is grouped into a Medicare Part A and Part B classification system. Part A items such as room and board, radiology, physical therapy, and supplies provided to hospital patients are valued by multiplying the billed charge for each item by the cost center-specific Medicare cost-to-charge ratio for the year in which the service was delivered.
Medicare Part B reimbursement rates are applied to health care services that are covered under Medicare Part B, including physician-billed services (eg, consultation, diagnostic, and therapeutic procedures) and other services such as laboratory and radiology.18 The Olmsted County Healthcare Expenditure and Utilization Database has cost data for medications provided during hospitalization, but medications dispensed to outpatients are not available in this database and the associated costs thus cannot be valued. Similarly, because it captures only billed costs, the value of postoperative convalescence is not captured. Although the information on health care service use reflects the practice patterns at the Mayo Clinic, the value of each unit of service has been adjusted to national norms by using widely accepted valuation techniques.20
All-cause cost from the day of index admission through the end of the sixth week was extracted for the study patients. Cost data were analyzed both descriptively and after multivariable adjustment. For the matched data, mean (standard deviation) and median costs were provided for robotic compared with abdominal hysterectomy and for robotic compared with vaginal hysterectomy.
Multivariable adjustment of costs in the matched samples was conducted using a generalized linear modeling framework.21 The generalized linear modeling methodology adjusted for predictors that significantly differed between groups, even after propensity score matching. Furthermore, it also included factors that were considered determinants of cost, including operative time, uterine weight, and binary indicators of intraoperative and postoperative complications. Intraoperative complications included incidental cystotomy, intraoperative transfusion, bowel injury, ureteral injury, vascular injury, and conversion to laparotomy in the robotic cases.
Occurrence of complications such as pulmonary edema, pulmonary embolism, and sepsis was captured with a binary indicator variable. Predicted costs for robotic and abdominal or vaginal procedures and the difference in predicted costs between groups were estimated using the recycled predictions method based on 200 bootstrap replications.22 We also assessed the average marginal effects of intraoperative and postoperative complications, which provided the estimated difference in total costs between those with and without complications (assuming all other patient characteristics remained the same).
Table 1 describes propensity score-matched baseline characteristics between robotic and abdominal hysterectomy and robotic compared with vaginal hysterectomy. Overall, baseline characteristics such as age, race, insurance status, body mass index, gravidity, live births, vaginal deliveries, and prior abdominal surgery were similar between abdominal and robotic hysterectomy cohorts. A higher proportion of patients with an endocrine comorbidity was noted in the abdominal group. The distribution of surgical indications was also significantly different between these groups (Table 1). No difference was seen between groups for the indications of pain or fibroids.
Baseline characteristics between robotic and vaginal approaches showed several significant differences, although comorbidities were similar. Age at surgery was higher in the robotic group. Distribution of race was different with a significantly higher proportion of nonwhite patients and a significantly lower proportion of patients of unknown race in the robotic group. Insurance status also differed between robotic and vaginal surgery groups with significantly fewer commercially insured patients and significantly more Medicare patients in the robotic cohort. Patients undergoing vaginal hysterectomy had an overall higher gravidity, number of live births, and number of vaginal deliveries than patients in the robotic group. Differences in indications were also noted with a higher proportion of patients undergoing vaginal hysterectomy for abnormal uterine bleeding and a higher proportion of patients undergoing robotic hysterectomy for gynecologic cancer. Indications such as hyperplasia, pain, fibroids, adnexal mass, and others were similar between the two groups (Table 1).
Table 2 shows intraoperative and postoperative characteristics, including complications, for robotic compared with abdominal and robotic compared with vaginal hysterectomy. When comparing the robotic and abdominal hysterectomy groups, intraoperative complications were significantly more frequent in abdominal hysterectomy, although the occurrence of postoperative complications and uterine weight did not differ between groups. Length of hospitalization was shorter (1.6 compared with 3.4 days; P<.001) and operative time was longer (3.0 compared with 1.8 hours; P<.001) for the robotic group compared with the abdominal hysterectomy group.
When comparing the robotic and vaginal hysterectomy cohorts, the robotic group had a higher rate of postoperative complications, a longer operative time, and an overall larger uterine weight. Length of hospitalization, however, was significantly shorter for the robotic hysterectomy group. Complications occurring for all routes are shown in Table 3.
Costs incurred in the first 6 weeks after the index surgery were not significantly different between the abdominal and robotic groups ($14,679 compared with $15,588; P=.35) (Table 4). In addition, costs were not significantly different between robotic and abdominal groups for the subsets of patients with no complications, with either an intraoperative or postoperative complication, or with both intraoperative and postoperative complications as indicated by the overlapping 95% confidence intervals in Table 5. The additional incurred cost of both intraoperative and postoperative complications compared with no complication was $11,430 and $12,137 for the robotic and abdominal groups, respectively, suggesting that surgical complications contribute a considerable amount to the overall cost of the procedure.
The robotic group incurred an adjusted cost that on average was $2,253 higher per patient than that of the vaginal hysterectomy group (Table 4). A similar cost difference ($2,154) between robotic and vaginal patients persisted for the subset of patients with no complications (Table 5). However, costs were not significantly different between robotic and vaginal hysterectomy groups for patients with either an intraoperative or postoperative complication or with complications both intraoperatively and postoperatively (Table 5). The additional incurred cost of both intraoperative and postoperative complications compared with no complication was $6,016 and $5,020 for the robotic and vaginal groups, respectively.
Our results reveal that robotic hysterectomy is more costly than vaginal hysterectomy but of similar cost to the abdominal approach. The cost difference between vaginal and robotic approaches persisted in patients who experienced no intraoperative or postoperative complications, but the difference was not significant in patients who had both complications. Patients undergoing vaginal hysterectomy had shorter operative times (1.2 compared with 3.1 hours) and a lower postoperative complication rate (8.0% compared with 17.9%) compared with patients who had a robotic procedure. Although the vaginal hysterectomy group had a longer hospital stay (1.9 compared with 1.5 days), cost differences between groups appeared to be driven by decreased operative times and lower complication rates in the vaginal hysterectomy group. Vaginal hysterectomy at our institution routinely uses an aggressive McCall culdoplasty for apical support, which is not typically performed with robotically assisted hysterectomy; this may have affected the difference seen in hospital stay.
One study comparing costs between robotic, abdominal, and vaginal approaches also included laparoscopic hysterectomy and examined 1,474 consecutive cases.13 Although cost differences were similar to our study, it included costs only from admission through discharge and did not consider costs resulting from complications or readmissions. Furthermore, nearly two-thirds of cases in that series were robotic; because they attempted to treat complex cases robotically, complexity may have accounted for most of the reported cost differences. Lastly, use of consecutive cases introduces sampling bias. The second study also included only inpatient charges and similarly used consecutive cases.14 Our approach to evaluation of cost mitigates these problems. We included all costs billed through 6 weeks after the procedure by using the Olmsted County Healthcare Expenditure and Utilization Database, which captures every dollar spent on medical care (eg, initial surgery, hospitalization, inpatient medications, subsequent physician visits, interventions for complications, readmissions). This all-cause approach provides an umbrella cost estimate for the different approaches and allows an overall cost comparison using nationally standardized dollar amounts. Selection bias was also decreased, although not eliminated, in our study by using comparison cohorts from the years immediately preceding the introduction of the robotic surgery at our institution and through propensity score matching.
In the 2 years preceding the introduction of robotic hysterectomy, 50.6% of the nearly 1,000 hysterectomies performed annually at our institution were vaginal. In 2011 and 2012, both the absolute number and the percentage of vaginal hysterectomy decreased to an average of 35.3%. Abdominal hysterectomy similarly decreased from 44.9% to 29.8% of the total hysterectomies performed during these years. Assuming a consistent rate (approximately 1,000 hysterectomies annually), 153 fewer vaginal hysterectomies were performed (converted to robotically assisted hysterectomy) in 2012 compared with the years immediately preceding the introduction of robotic surgery at an estimated annual expense of $344,709. If this trend is representative of the 600,000 hysterectomies performed annually in the United States, adoption of the robotic system represents an estimated increased cost of nearly $207 million per year.
Our study likely underestimates the cost differences between approaches. First, the analysis did not account for the initial purchase price of the robotic system, yearly maintenance costs, or decreased operating room use for robotic cases. Second, costs associated with the learning curve also were not considered; this may consist of up to 91 cases when using operative complications as the measure of proficiency.23 Third, the analysis was restricted to the use of Medicare-reimbursed amounts in the Olmsted County Healthcare Expenditure and Utilization Database. If true billing costs were used, including those from private insurers, costs for all approaches may be considerably higher and cost differences may also vary. Moreover, we assigned an average expense cost for variable expenses in robotic surgery, but these have been shown to be a major source of cost12 regardless of the specific amount used in a particular procedure. Our group has recently shown that a novel approach to the perioperative management of patients having vaginal reconstructive procedures resulted in a significantly shorter hospital stay and decreased hospitalization cost.24 It is therefore plausible that the widespread implementation of this approach may further limit the duration of hospitalization in patients having vaginal hysterectomy and amplify the cost difference observed.
The inclusion of costs in the postoperative period, although advantageous from an all-cause standpoint, may potentially capture additional costs not associated with the procedure itself because hospital costs and physician visits unrelated to the procedure may have been included. However, the cost of robotic procedure remained higher than that of vaginal surgery, even among patients without complications, suggesting that capture of unrelated costs likely did not have a major effect in our analysis. Furthermore, the Olmsted County Healthcare Expenditure and Utilization Database captures the cost of pharmaceuticals provided in the inpatient setting only (ie, no outpatient pharmaceutical data). Operative times for hysterectomy routes are major drivers of cost. Differences in operative times at other institutions may affect the external validity of our findings, although operative times were similar and, in some cases, less than that of other studies.14,25 We were also unable to assess the cost of total laparoscopic hysterectomy compared with robotic hysterectomy because the laparoscopic approach is infrequently performed at our institution. We acknowledge that the relatively low number of patients is a limitation of the study. However, the power analysis (see Appendices 1–4, available online at http://links.lww.com/AOG/A464) shows that the study was adequately powered to detect the differences in outcomes reported.
Based on these data, we conclude that the vaginal approach is safer, faster, and more cost-effective than the robotic approach. Although robotic hysterectomy may have advantages in certain niches in gynecologic practice, the data presented here caution against the widespread use of this technology. Rather, the data reinforce the American College of Obstetricians and Gynecologists’ recommendation that the preferred route of hysterectomy be vaginal.3,4
1. Bloche MG. Beyond the ‘R word’? Medicine's new frugality. N Engl J Med 2012;366:1951–3.
2. Whiteman MK, Hillis SD, Jamieson DJ, Morrow B, Podgornik MN, Brett KM, et al.. Inpatient hysterectomy surveillance in the United States, 2000–2004. Am J Obstet Gynecol 2008;198:34.e1–7.
3. Choosing the route of hysterectomy for benign disease. ACOG Committee Opinion No. 444. American College of Obstetricians and Gynecologists. Obstet Gynecol 2009;114:1156–8.
5. Dorsey JH, Holtz PM, Griffiths RI, McGrath MM, Steinberg EP. Costs and charges associated with three alternative techniques of hysterectomy. N Engl J Med 1996;335:476–82.
6. Sculpher M, Manca A, Abbott J, Fountain J, Mason S, Garry R. Cost effectiveness analysis of laparoscopic hysterectomy compared with standard hysterectomy: results from a randomised trial. BMJ 2004;328:134.
7. Warren L, Ladapo JA, Borah BJ, Gunnarsson CL. Open abdominal versus laparoscopic and vaginal hysterectomy: analysis of a large United States payer measuring quality and cost of care. J Minim Invasive Gynecol 2009;16:581–8.
8. Advincula AP, Xu X, Goudeau S 4th, Ransom SB. Robot-assisted laparoscopic myomectomy versus abdominal myomectomy: a comparison of short-term surgical outcomes and immediate costs. J Minim Invasive Gynecol 2007;14:698–705.
9. Pasic RP, Rizzo JA, Fang H, Ross S, Moore M, Gunnarsson C. Comparing robot-assisted with conventional laparoscopic hysterectomy: impact on cost and clinical outcomes. J Minim Invasive Gynecol 2010;17:730–8.
10. Barnett JC, Judd JP, Wu JM, Scales CD Jr, Myers ER, Havrilesky LJ. Cost comparison among robotic, laparoscopic, and open hysterectomy for endometrial cancer. Obstet Gynecol 2010;116:685–93.
11. Patel M, O'Sullivan D, Tulikangas PK. A comparison of costs for abdominal, laparoscopic, and robot-assisted sacral colpopexy. Int Urogynecol J Pelvic Floor Dysfunct 2009;20:223–8.
12. Wright JD, Ananth CV, Lewin SN, Burke WM, Lu YS, Neugut AI, et al.. Robotically assisted vs laparoscopic hysterectomy among women with benign gynecologic disease. JAMA 2013;309:689–98.
13. Landeen LB, Bell MC, Hubert HB, Bennis LY, Knutsen-Larson SS, Seshadri-Kreaden U. Clinical and cost comparisons for hysterectomy via abdominal, standard laparoscopic, vaginal and robot-assisted approaches. S D Med 2011;64:197–9.
14. Wright KN, Jonsdottir GM, Jorgensen S, Shah N, Einarsson JI. Costs and outcomes of abdominal, vaginal, laparoscopic and robotic hysterectomies. JSLS 2012;16:519–24.
15. Heisler CA, Weaver AL, Melton LJ 3rd, Gebhart JB. Effect of additional reconstructive surgery on perioperative and postoperative morbidity in women undergoing vaginal hysterectomy. Obstet Gynecol 2009;114:720–6.
16. Dindo D, Demartines N, Clavien PA. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 2004;240:205–13.
17. D'Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265–81.
18. Rocca WA, Yawn BP, St Sauver JL, Grossardt BR, Melton LJ 3rd. History of the Rochester Epidemiology Project: half a century of medical records linkage in a US population. Mayo Clin Proc 2012;87:1202–13.
19. St Sauver JL, Grossardt BR, Yawn BP, Melton LJ 3rd, Pankratz JJ, Brue SM, et al.. Data resource profile: the Rochester Epidemiology Project (REP) medical records-linkage system. Int J Epidemiol 2012;41:1614–24.
20. Lipscomb J, Ancukiewicz M, Parmigiani G, Hasselblad V, Samsa G, Matchar DB. Predicting the cost of illness: a comparison of alternative models applied to stroke. Med Decis Making 1998;18(suppl):S39–56.
21. Blough DK, Madden CW, Hornbrook MC. Modeling risk using generalized linear models. J Health Econ 1999;18:153–71.
22. Basu A, Rathouz PJ. Estimating marginal and incremental effects on health outcomes using flexible link and variance function models. Biostatistics 2005;6:93–109.
23. Woelk JL, Casiano ER, Weaver AL, Gostout BS, Trabuco EC, Gebhart JB. The learning curve of robotic hysterectomy. Obstet Gynecol 2013;121:87–95.
24. Kalogera E, Bakkum-Gamez JN, Jankowski CJ, Trabuco E, Lovely JK, Dhanorker S, et al.. Enhanced recovery in gynecologic surgery. Obstet Gynecol 2013;122:319–28.
© 2014 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
25. Kho RM, Hilger WS, Hentz JG, Magtibay PM, Magrina JF. Robotic hysterectomy: technique and initial outcomes. Am J Obstet Gynecol 2007;197:113.e1–4.