Variation in prostate surgery costs and outcomes in the USA: robot-assisted versus open radical prostatectomy

Given the rapid adoption of RARP coupled with the absence of any large randomized controlled trial comparing RARP and ORP, there was a trend to compare the rate of perioperative complications, blood transfusion, requirement of additional cancer therapy, positive surgical margin (PSM), length of stay (LOS) and cost of treatment between the RARP and ORP. A single center Phase III randomized controlled trial study, conducted between 2010 and 2014, found no significant difference in functional (urinary function, sexual function), oncological outcomes (PSM status, biochemical and imaging evidence of progression at 24 months) and postoperative complications between RARP and ORP [8]. No consistent evidence was reported by multiple observational studies as well. Observational studies with smaller number of patients comparing ORP and minimum-invasive RP (MIRP) reported no significant difference in perioperative complications [9,10]. A systematic review of ORP, LRP and RARP reported similar perioperative complication rate between ORP and RARP and significantly lower blood loss and transfusion rate for patients undergoing RARP [11]. Large scale observational studies using 2003–2005 SEER-Medicare data reported no significant difference in postoperative general medical or surgical complication rate within 90 days, genitourinary and bowel complication within 365 days, and receipt of subsequent cancer therapy between ORP and MIRP [12]. Analysis of 2003–2007 SEER Medicare data found less respiratory and miscellaneous surgical complications but higher postoperative genitourinary complications with MIRP compared with ORP [13]. Contrary to SEER-Medicare data results, a study using the 2005–2010 National Surgery Quality Improvement Program data reported significantly less rate of cardiovascular, deep vein thrombosis, pulmonary embolism complication, surgical site infection and units of blood transfusion for MIRP compared with ORP [14].

In 2008, the US FDA approved a robot-assisted modifier code (International Classification of Disease, Ninth Revision [ICD-9] clinical modification procedure code 17.42) to identify accurately robot-assisted procedures. A study analyzing the 2008–2009 Nationwide Inpatient Sample (NIS) data and utilizing the robot-assisted modifier code concluded that RARP was superior to ORP in terms of intraoperative, postoperative complications and blood transfusions. Patients with RARP compared with ORP had significantly lower probability of cardiac (Odds Ratio [OR]: 0.73; 0.53–0.99), respiratory (OR: 0.54; 0.42–0.69) and vascular (OR: 0.59; 0.37–0.95) complications [15]. Another study utilizing the 2008–2009 SEER-Medicare data found similar risk of overall postoperative complications during first 30 day and 90 day after surgery and readmission rates across ORP and RARP. The RARP was associated with an elevated risk of genitourinary and miscellaneous medical complications during first 30 day and 90 day postsurgery but with a lower risk of receiving a blood transfusion [16]. Analysis of 2003–2012 SEER-Medicare data reported that the risk of receiving additional radiation and androgen deprivation therapy was significantly lower for RARP (22%) compared with ORP with no significant difference in PCa-specific and all-cause mortality between the two procedures over a median follow-up of 6.5 years [17]. Another study utilizing the 2003–2013 Premier Hospital database reported that patients undergoing RARP had 32% less complications during the first 90 day postsurgery, and received less blood products compared with those undergoing ORP. However, no difference (OR: 1.00; 0.73–1.38) in the occurrence of major complications (Clavien grade 3–5) was reported between the two procedures [6].

When PSM was compared across ORP, RARP and LRP, no conclusive evidence of superiority in favor of any one procedure was observed [18–22]. However, MIRP was associated with shorter hospital LOS compared with ORP [6,12,13,15,16,23]. Studies further stated that patients undergoing RARP could benefit from earlier return to work compared with patients undergoing ORP [6]. The cost of treatment with RARP was found to be higher than that of ORP [6,16,24]. A single institution US study of 643 patients reported that RARP had higher cost even after excluding the purchase cost and annual maintenance cost of the robot compared with ORP. The median direct cost (2007 estimate) was $6752, $5687 and $4437 for RARP, LRP and ORP, respectively [24]. Another study conducted in 2008–2009 reported that median cost of first-year postsurgery was approximately $1400 more for RARP compared with ORP among Medicare patients who were treated with RP [16]. Using the Premier Hospital database, a study reported that during 2011–2013, the adjusted 90 day direct hospitalization costs were significantly higher ($2313; 95% CI: $106–$4521) for RARP compared with ORP. However, this cost estimation did not include the fixed cost of purchasing the robot and the maintenance cost, which possibly underestimated the cost differences between two procedures [6]. Some researchers, however, argue that RARP may be more costly upfront, but patients treated with RARP may need less additional cancer therapy later on, which could possibly offset the high initial cost [17].

In essence, though RARP is a costly procedure compared with ORP, it is superiority over ORP in different aspects of perioperative outcomes seems largely inconclusive. The 2017 American Urological Association (AUA)/American Society for Radiation Oncology (ASTRO)/Society of Urologic Oncology (SUO) guidelines recommended that clinicians should inform patients with localized PCa that cancer control, continence and sexual recovery outcomes are similar between ORP and RARP [25]. A few observational large-scale studies utilizing SEER-Medicare data have evaluated the clinical and economic outcomes of RP; however, the results were generalizable only to elderly population. Similarly, studies analyzing the Premier Hospital database had limitations because the participating hospitals were predominantly small-to-medium sized, nonteaching type and serving mostly urban population, whereas a significant portion of patients with PCa in the USA are treated in academic hospitals and cancer treatment centers [26,27]. Although the American Hospital Association database has 37.82% of hospitals listed as teaching hospitals, the Premier Hospital database has only 26.60% as teaching hospital (2016) [28].

Thus, there is a need to utilize a large, nationally representative hospitalization data to compare different outcomes between RARP and ORP. This study utilizes data from the NIS of Healthcare Cost and Utilization Project (HCUP) from January 2010 to September 2015 (latest available year of data) to compare the perioperative complications, inpatient cost, and LOS between RARP and ORP.

The primary objective of the study was to compare perioperative complications, inpatient cost and LOS among adult patients with PCa who had undergone RP using either RARP or ORP. A secondary objective was to compare the inpatient cost between RARP and ORP across different sociodemographic, clinical and hospital characteristics.

A total of 89,972 discharge records with a primary diagnosis of PCa were identified from 2010 to 2015. Records with age less than 18 years (n = 77) and with any other metastatic cancer (n = 3128) were excluded from the final analyses. Among the remaining 86,767 discharge records, 69,414 records had RP as a procedure listed during their stay. Further, 405 records were excluded as those included nonrobotic LRP. Finally, 69,009 records were included in analysis among which 52,151 (75.57%) had RARP and 16,858 (24.43%) had ORP (Figure 1). A statistically significant (p < 0.0001) upward trend was observed in RARP from 2010 (66.69%) to 2015 (83.27%).

This study found that although the total number of RP has decreased since 2011, the proportion of RARP has increased steadily from 2010 to 2015. This was consistent with a study of NIS data from 1998 to 2011 which reported a 7% decline in the total number of RP conducted in the USA [33]. The increase in the proportion of RARP was also consistent with those reported by other studies [6].

The odds of any complication was significantly lower for RARP compared with ORP. Adjusted multivariate analysis found RARP as a favorable technique over ORP for each of the specific complications except vascular complication. The lower odds of complications with RARP was also observed in other studies [34]. Although previous studies have reported RARP to have less odds of producing only cardiac, respiratory and vascular complications compared with ORP, or miscellaneous surgical complications, and increased risk of genitourinary complications; more recent studies have reported an overall lower morbidity with RARP compared with ORP [6,13,15]. The present study found results consistent with recently published studies and further provides evidence of the benefit of RARP over ORP regarding perioperative outcomes. The favorable outcomes with RARP could possibly be attributed to rapid adoption of the surgical technique over the years, the increased experience of surgeons as they performed more RARP procedure, advancement in technology and possibly better ergonomics for surgeons [34–37]. A previous study also observed an improvement in prostatectomy-related outcomes such as continence, potency recovery and presence of surgical margins, which was attributed to the increased experience of surgeon and advancement of the surgical technique [38]. A recently published study of 412 patients who underwent RARP between 2014 and 2016 in Australia also reported high level of patient satisfaction with RARP. Majority of the patients mentioned that they would recommend RARP as a procedure to others [39].

The increased benefits of RARP over ORP, however, come at an increased cost. Adjusted mean inpatient cost for RARP was significantly higher compared with ORP by almost $3000. However, the difference in the mean cost between the two procedures diminished by almost 38% over 6 years from 2010 to 2015, which was close to a previously reported estimate of 45% decrease from 2008 to 2010 until 2013 [6].

The proportion of people receiving RARP increased with higher household income. The mean cost of RARP and the difference in mean cost between RARP and ORP was highest for black men, which was consistent with a previous study [40]. A study using 2004–2011 SEER data reported that African–American men were less likely to receive definite therapy for PCa compared with white men [41]. The present study found mean cost of RARP for blacks was $1703 more compared with whites, which might act as an economic barrier for black men to access RARP and thus, contributes to the disparity in the receipt of PCa treatment. This study found that among all races, black men had the lowest proportion of receipt of RARP. The cost of RARP increased with the number of comorbidities, but the difference in mean cost between RARP and ORP decreased with the increase in number of comorbidities. Significant sociodemographic variation in the receipt of RARP and cost of RARP could act as a deterrent to receive appropriate care among PCa patients in specific groups of population.

The cost of both RARP and ORP was lower in urban teaching hospitals than nonteaching hospitals, rural hospitals and in hospitals with high compared with low volume of RP. This finding was similar to other studies using 2009–2011 NIS data or Premier Hospital database [40,42]. The present study further found that compared with private invest-own hospitals, the cost of RARP and ORP was higher in government nonfederal and private nonprofit hospitals, which was similar to the findings by Faiena et al. [42]. Although their study compared the total cost of RP using only 2009–2011 NIS data, we separately compared the cost of both RARP and ORP using latest available data up to 2015. However, private invest-own hospitals accounted for only 9.15% of the total number of RP and 8.96% of total RARP. The very low proportion of RARP in private invest-own hospitals may be due to high initial investment cost to acquire robotic system and high annual maintenance cost.

This study also observed a regional variation in the cost of RARP. The cost of RARP was lowest in the northeast, and highest in the west region. The difference in cost between RARP and ORP was highest in west followed by the midwest region. The cost of RARP was significantly higher than ORP in all regions except northeast, which was in congruence with a previous study [42]. The NIS dataset does not provide any opportunity to explore reasons behind low cost of RARP compared with ORP in the northeast region. The present study found that adjusted LOS was smaller for RARP compared with ORP by 0.85 days, which was similar to the study by Leow et al. [6]. The shorter LOS might contribute to consumption of less hospital resources like room and board for patients, who went through RARP.

The strength of this study is in using a nationally representative discharge records, which includes all payers and uninsured population. Contrary to SEER Medicare data, usage of NIS dataset will not restrict the finding of this study only in elderly population. Adequate representation of urban teaching and large hospitals reflect the nature of hospitals in which many patients with PCa are treated in USA, which was not captured in Premier Hospital database.

Analyzing nationally representative discharge records, the present study found that men undergoing RARP experienced fewer peri-operative outcomes but the cost was higher compared with ORP. Additionally, racial and regional differences were observed in the utilization of these procedures, which could possibly act as a barrier to their utilization resulting in healthcare disparity. In future, it would be also interesting to measure the indirect costs due to productivity losses for these procedures among patients, which can be utilized to value these procedures from a societal perspective.

K Mukherjee performed conception, design of the work, acquisition, analysis, interpretation of data for the work, drafting the work or revising it critically for important intellectual content. KM Kamal performed design of the work, interpretation of data for the work, drafting the work or revising it critically for important intellectual content.

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.

The authors state that the HCUP-NIS public use unidentified dataset does not require any IRB approval. However, the authors completed the HCUP Data Use Agreement Training for acquiring and using the data.