Impact of Caseload on Total Hospital Charges: A Direct

Comparison Between Minimally Invasive and Open Radical

Prostatectomy—A Population Based Study

Firas Abdollah,* Lars Budäus,* Maxine Sun, Monica Morgan, Rupinder Johal,Rodolphe Thuret, Kevin C. Zorn, Hendrik Isbarn, Alexander Haese, Claudio Jeldres,Paul Perrotte, Francesco Montorsi, Markus Graefen and Pierre I. Karakiewicz†From the Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center (FA, LB, MS, MM, RJ, RT, KCZ, CJ, PP, PIK),Montreal, Quebec, Canada, Vita-Salute University/Hospital San Raffaele (FA, FM), Milan, Italy, and Martiniclinic, Prostate Cancer CenterUniversity Hospital Hamburg-Eppendorf (LB, HI, AH, MG), Hamburg, Germany

Purpose: We tested the relationship between caseload and total hospital chargesafter stratifying by minimally invasive and open radical prostatectomy.Materials and Methods: We evaluated 1,188 vs 3,354 men treated with mini-mally invasive vs open radical prostatectomy in the Florida Hospital Inpatientsdata file in 2008. Caseload was defined as the count of procedures performed byeach surgeon between the study start on January 1, 2008 and the date of eachprocedure. Patients were divided into tertiles based on their procedure specificcaseload. Univariate and multivariate analysis was done to address the relationbetween caseload and total hospital charges for the minimally invasive and openprocedures. Covariates were patient age, race, comorbidity, and length of stay.Results: Median total hospital charges for minimally invasive and open radicalprostatectomy were $33,234 and $33,674, respectively (p � 0.03). Median totalhospital charges in the low, intermediate and high minimally invasive vs openprocedure caseload tertiles were $41,765, $34,799 and $28,780 vs $35,642,$34,726 and $32,726, respectively. On multivariate analysis with the high min-imally invasive caseload tertile as the reference category the increments of theprobability of charges in excess of the 2008 median of $33,588 were 3.9 and8.1-fold for the intermediate and low caseload minimally invasive procedures,and 2.5, 3.6 and 2.8-fold for the high, intermediate and low caseload openprocedures, respectively (each p �0.001).Conclusions: Overall median total hospital charges are virtually the same forminimally invasive and open radical prostatectomy. However, total hospitalcharges for the minimally invasive procedure have a more sensitive caseloadeffect, as evidenced by the wider distribution of the median of minimally invasivecaseload specific total hospital charges vs that of open radical prostatectomy. Thehigh caseload minimally invasive procedure resulted in the lowest total hospitalcharges relative to all other minimally invasive and open radical prostatectomycategories.

Key Words: prostate; prostatic neoplasms; prostatectomy;

Abbreviations

and Acronyms

CCI � Charlson comorbidity index

LRP � laparoscopic RP

MIRP � minimally invasive RP

ORP � open RP

RP � radical prostatectomy

Submitted for publication June 14, 2010.Study received institutional review board ap-

proval.Supported by German Federal Ministry of Ed-

ucation and Science in the framework of theprogram for medical genome research FKZ:01GS08189 (LB) and the University of MontrealHealth Center Fonds de la Recherche en Santé duQuebec, University of Montreal Department ofSurgery and University of Montreal Health CenterFoundation (PIK).

* Equal study contribution.† Correspondence: Cancer Prognostics and

Health Outcome Unit, University of MontrealHealth Center, 1058, rue St-Denis, Montreal,Quebec, Canada, H2X 3J4 (telephone: 514-890-8000-35336; FAX: 514-227-5103; e-mail: pierre.karakiewicz@unmontreal.ca).

surgical procedures, minimally invasive; hospital charges

RADICAL prostatectomy represents thestandard of care for patients with

clinically localized prostate cancer

0022-5347/11/1853-0855/0THE JOURNAL OF UROLOGY®

© 2011 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION AND RES

and life expectancy greater than 10years.1,2 New technologies such as

MIRP are increasing in popularity

Vol. 185, 855-861, March 2011Printed in U.S.A.

EARCH, INC. DOI:10.1016/j.juro.2010.10.051www.jurology.com 855

IMPACT OF CASELOAD ON PROSTATECTOMY TOTAL HOSPITAL CHARGES856

while the utilization rate of traditional ORP is de-creasing.3–7 Nonetheless, novel techniques are cus-tomarily compared to ORP. Numerous comparisonshave been made between MIRP and ORP showingthat MIRP has several potential advantages, includ-ing lower blood loss and transfusion rate, decreasedpostoperative pain and shorter hospital stay.8–10

However, higher cost may represent a potentialMIRP disadvantage, as suggested by 2 institutionalanalyses.11,12

To our knowledge no group to date has tested theeffect of caseload on cost, especially in a large, con-temporary, population based cohort. To address thelack of such data we postulated that there may beimportant differences between MIRP and ORP totalhospital charges. Moreover, we hypothesized thattotal hospital charges may be sensitive to the effectof surgical expertise, defined as caseload. Specifi-cally higher caseload may exert a protective effect ontotal hospital charges. Based on greater technicalcomplexity and on its novelty relative to ORP wealso postulated that MIRP total hospital chargesmay be more sensitive to the effect of caseload thanORP total hospital charges. To provide the mostcontemporary estimates we focused on individualsexclusively operated on in 2008 who had records inthe Florida Hospital Inpatient Datafile, which al-lows the ascertainment of virtually all RPs in thatstate.

METHODS AND MATERIALS

Study PopulationWe relied on 2008 Florida Hospital Inpatient Datafile,which provides inpatient medical record information fromadmission until discharge. Data collection at participatinginstitutions is a mandatory process done by trained pro-fessional coders. Before being made available to interestedparties, data are audited for errors.13 Analyzed elementsconsisted of diagnostic and procedure codes based on ICD-9-CM,4 and encrypted patient and surgeon identifiers.Patient characteristics consisted of age at admission(years), race (white vs other), comorbidity (diagnosticcode), length of stay (days) and total hospital charges (inUnited States dollars).

All patients with a primary diagnosis of prostate cancer(ICD-9-CM code 185) were considered for analysis. The RPprocedure code (ICD-9-CM 60.5) identified 4,542 patientswith RP who represented the focus of the study. Of thesepatients 1,188 were further classified as having under-gone MIRP using the minimally invasive modifier code(ICD-9-CM 54.21). The remaining 3,354 patients wereconsidered to have undergone ORP. Baseline comorbiditystatus was quantified with CCI according to ICD-9-CMdiagnostic codes.15

Statistical AnalysisDescriptive statistics consisted of tests of means, medians

and proportions. Total hospital charges are shown as the

total undiscounted amount of dollars charged to patientsfor all services rendered at the hospital from admission todischarge, referred to as charges. Focus was directed to-ward the relationship between the type of RP (MIRP vsORP) and charges reported for each admission. We as-sessed the association between charges and patient case-load, defined as the count of procedures performed by theoperating surgeon between the study start and the date ofMIRP or ORP in each patient. Caseload count was proce-dure specific. Patients were divided into tertiles based ontheir procedure specific caseload, as in previously reportedmethodology.16,17

We used logistic regression models to quantify the re-lation between procedure specific caseload tertiles andcharges in excess of the overall median value. Adjustmentwas made for age, race, CCI (stratified in 4 categories as 0to 3 or more) and length of stay. To examine the potentialeffect of outliers we repeated the mentioned methodologyin a subanalysis excluding 454 patients with charges be-low and above the 5th and 95th percentiles, respectively.

All analysis was done using SPSS®, version 15.0 with2-sided p �0.05 considered statistically significant. Whenappropriate, we applied the Bonferroni adjustment18 formultiple comparisons.

RESULTS

In 2008 in Florida 1,188 MIRPs and 3,354 ORPswere performed (table 1). Patients with MIRP weremore often white than those with ORP (83.5% vs79.7%, p �0.001). Average length of stay for MIRPand ORP was 1.7 (median 1.0, range 1 to 28) and 2.1days (median 2.0, range 1 to 50), respectively(p �0.001). No differences existed between age andCCI in patients with MIRP and ORP. Low, interme-diate and high tertiles corresponded to caseloadstrata of 1 to 6, 7 to 25 and 26 to 166 for MIRP vs 1to 5, 6 to 25 and 26 to 384 for ORP, respectively.Patient caseload tertiles indicated that 61%, 29%and 10% of surgeons were in the low, intermediateand high tertiles for MIRP vs 56%, 35% and 9% forORP, respectively (fig. 1).

Median charges were $33,234 (mean $41,588,range $5,526 to $272,759) for MIRP vs $33,674(mean $43,168, range $3,971 to $612,403) for ORP(p � 0.03). When stratified by procedure specificcaseload tertiles (fig. 2), median charges in thelow, intermediate and high tertiles were $41,765,$34,799 and $28,780 for MIRP vs $35,642, $34,726and $32,726 for ORP, respectively. After adjustingfor multiple comparisons median charges were sig-nificantly lower for MIRP than for ORP between thehigh caseload categories (p �0.001). Median chargeswere significantly higher for MIRP than for ORPbetween the low caseload categories (p �0.001). Nostatistically significant differences were recorded be-tween the intermediate caseload categories after ad-justing for multiple comparisons (p � 0.04 vs

�0.016).

Table 1. General characteristics and descriptive statistics of 4,542 men who underwent RP in 2008 in Florida by surgery type and procedure specific caseload tertile

Overall Total MIRP

MIRP Caseload Tertiles

Total ORP

ORP Caseload Tertiles

p ValueLow Intermediate High p Value Low Intermediate High p Value

No. pts (%) 4,542 (100) 1,188 (26.2) 403 391 394 3,354 (73.8) 1,059 1,186 1,109Age: 0.10 0.47 0.90

Mean 61 61 61 61 62 61 62 61 61Median 62 62 62 61 63 62 62 62 61Range 30–90 34–90 34–90 42–79 41–78 30–81 40–80 38–80 37–81

No. race (%): �0.001 0.13 �0.001White 3,618 (79.7) 992 (83.5) 321 (79.7) 310 (79.3) 361 (91.6) 2,626 (78.3) 814 (76.9) 922 (77.7) 890 (80.3)Others 924 (20.3) 196 (16.5) 82 (20.3) 81 (20.7) 33 (8.4) 728 (21.7) 245 (24.5) 264 (22.3) 219 (19.7)

No. CCI (%): 0.71 �0.001 0.120 3,582 (78.9) 945 (79.5) 314 (77.9) 319 (81.6) 312 (79.2) 2,637 (78.6) 798 (75.4) 919 (77.5) 920 (83.0)1 217 (4.8) 42 (3.5) 14 (3.5) 12 (3.1) 16 (4.1) 175 (5.2) 64 (6.0) 61 (5.1) 50 (4.5)2 549 (12.1) 150 (12.6) 59 (14.6) 45 (11.5) 46 (11.7) 399 (11.9) 139 (13.1) 161 (13.6) 99 (8.9)3 or Greater 194 (4.3) 51 (4.3) 16 (4.0) 15 (3.8) 20 (5.1) 143 (4.3) 58 (5.5) 45 (3.8) 40 (3.6)

Caseload: — — — — — — — — 0.87Mean 37 31 40Median 12 12 12Range 1–384 1–166 1–384

Length of stay (days): �0.001 �0.001 �0.001Mean 2.0 1.7 2.1 1.6 1.4 2.13 2.9 2.0 1.5Median 1.0 1.0 2.0 1.0 1.0 2.00 2.0 2.0 1.0Range 1–50 1–28 1–28 1–28 1–18 1–50 1–50 1–22 1–22

Total hospital charges ($): �0.001 �0.001 0.03Mean 39,153 41,588 50,776 41,113 32,662 38,291 43,168 37,753 34,209Median 33,588 33,234 41,765 34,799 28,780 33,674 35,642 34,726 32,726Range 3,971–612,403 5,526–272,759 10,189–272,759 5,526–250,104 5,526–202,325 3,971–612,403 10,667–612,403 6,484–418,667 3,971–195,821

IMP

AC

TO

FC

AS

ELO

AD

ON

PR

OS

TATE

CTO

MY

TOTA

LH

OS

PITA

LC

HA

RG

ES

85

7

IMPACT OF CASELOAD ON PROSTATECTOMY TOTAL HOSPITAL CHARGES858

On multivariate logistic regression models ad-justed for age, race, CCI and length of stay MIRP inthe high caseload tertile was considered the refer-ence category (table 2). Increments in the probabil-ity of charges above the median of $33,588 were 8.1and 3.9-fold for MIRP in the low and intermediatecaseload tertiles, respectively (each p �0.001). Sim-ilarly increments in the probability of charges abovethe median were 2.5, 3.6 and 2.8-fold for ORP in thelow, intermediate and high caseload tertiles, respec-tively (all p �0.001).

Figure 1. MIRP and ORP surgeons by procedure specific low, indatabase.

Figure 2. Median total hospital charges in United States dollars f

intermediate and high patient caseload tertiles.

On subanalysis excluding patients with chargesbelow and above the 5th and 95th percentiles, re-spectively, overall median charges were $33,576(range $19,906 to $78,714). Median MIRP and ORPcharges were $31,854 (range $ 20,024 to $77,716)and $33,864 (range $19,908 to $78,714), respec-tively. Similar results were noted on this subanaly-sis (data not shown). Specifically on multivariateanalysis increments in the probability of chargesabove the median were 7.6 and 3.8-fold (eachp �0.001) for MIRP in the low and intermediate

iate and high patient caseload tertiles in 2008 Florida Inpatient

2 men who underwent MIRP and ORP in 2008 in Florida by low,

termed

or 4,54

IMPACT OF CASELOAD ON PROSTATECTOMY TOTAL HOSPITAL CHARGES 859

caseload tertiles, respectively, vs MIRP in the highcaseload tertile. Similarly increments in the proba-bility of charges above the median were 3.2, 4.5 and2.6-fold for ORP in the low, intermediate and highcaseload tertiles, respectively, vs MIRP in the highcaseload tertile (each p �0.001).

DISCUSSION

The popularity of MIRP is growing.3–7 However, thebelief that MIRP contributes to higher costs thanORP may undermine some of this growth. To datethis belief has been supported by factual data indi-cating higher upfront MIRP cost relative to ORP.11

Despite these data no group to our knowledge hasexamined the influence of caseload on MIRP costs orcharges. To address this limitation we examined therelationship between caseload and charges in a co-hort of men treated with MIRP or ORP in Florida.We exclusively focused on the most contemporaryyear, namely 2008, to provide the most up-to-dateestimates and exclude as much as possible the effectof the MIRP learning curve.

Results indicate that overall median charges forMIRP were virtually the same as for ORP ($33,234and $33,674, respectively). After stratifying mediancharges by procedure specific caseload tertiles wenoted that low caseload tertile MIRP was morecostly than low caseload tertile ORP ($41,765 vs$35,642). No difference in median charges was notedfor intermediate caseload tertile MIRP and ORP($34,799 and $34,726, respectively). Finally, highcaseload tertile MIRP resulted in lower mediancharges than high caseload ORP ($28,780 vs

Table 2. Univariate and multivariate logistic regression analysiunderwent RP in 2008 in Florida

Univa

OR (95% CI)

Procedure specific caseload tertiles:High MIRP 1.0 (referent)Intermediate MIRP 4.2 (3.1–5.8)Low MIRP 9.7 (7.01–13.4)High ORP 2.6 (1.99–3.4)Intermediate ORP 4.5 (3.41–5.8)Low ORP 4.4 (3.38–5.8)

Age 0.9 (0.9–0.9)Race:

White 1.0 (referent)Other 1.8 (1.5–2.1)

CCI:0 1.0 (referent)1 0.9 (0.7–1.2)2 1.3 (1.1–1.6)3 or Greater 1.7 (1.2–2.2)

Length of stay 1.5 (1.4–1.6)

$32,726). Multivariate analysis supported these uni-

variate observations. In the low and intermediatecaseload MIRP tertiles the odds of charges beingabove the median were 8.1 and 3.9-fold higher, re-spectively, than in the high caseload MIRP tertile.In the low, intermediate and high caseload ORPtertiles the odds of charges being above the medianwere 2.5, 3.6 and 2.8-fold higher, respectively, thanin the high caseload MIRP tertile. Similar resultswere observed on subanalysis excluding potentialoutliers.

Together our findings indicate that when low case-load surgeons performed MIRP, the procedure costmore than ORP performed by low caseload surgeons.The 2 low caseload categories contributed to sub-stantially higher charges than the intermediate orhigh caseload category regardless of RP type.When intermediate caseload surgeons were exam-ined, charges were higher than those of high case-load MIRP or ORP surgeons. However, there wasno meaningful difference between MIRP and ORPcharges for intermediate caseload surgeons. Fi-nally, when high caseload surgeons performedMIRP, the lowest charges were recorded. Chargesfor high caseload MIRP were lower than for highcaseload ORP.

These findings indicate that when a procedure isdone by an expert surgeon, it costs less than whendone by a less expert surgeon. This relation appliesregardless of RP type. Although previous literaturesuggests a cost disadvantage for MIRP relative toORP,11,12 our results show that this is not applicableto high caseload MIRP. A cost advantage existedwhen high caseload MIRP was compared to highcaseload ORP, suggesting that high caseload MIRP

icting total hospital charges above median in 4,542 men who

Multivariate

p Value OR (95% CI) p Value

— 1.0 (referent) —�0.001 3.9 (2.9–5.5) �0.001�0.001 8.1 (5.8–11.3) �0.001�0.001 2.5 (1.9–3.3) �0.001�0.001 3.6 (2.7–4.7) �0.001�0.001 2.8 (2.1–3.7) �0.001

0.01 0.9 (0.9–1.1) 0.08

— 1.0 (referent) —�0.001 1.5 (1.3–1.7) �0.001

— 1.0 (referent) —0.61 0.7 (0.6–1.0) 0.070.005 1.1 (0.9–1.3) 0.30.001 1.3 (0.9–1.7) 0.1

�0.001 1.4 (1.3–1.5) �0.001

s pred

riate

surgeons develop and practice in an environment

IMPACT OF CASELOAD ON PROSTATECTOMY TOTAL HOSPITAL CHARGES860

that is more cost conscious than that of high case-load ORP surgeons. This environment may consistof better defined clinical care pathways and an en-vironment that is more sensitive to details that pro-tect from escalating cost. Regardless of the underly-ing processes our findings are of major importance toinstitutional medical decision makers. It appearsthat from the charges perspective the highest sav-ings could be obtained by relegating patients fromlow and intermediate caseload surgeons to high case-load MIRP surgeons.

Charges in the intermediate caseload categorieswere virtually equal for MIRP and ORP. This find-ing is equally notable since it emphasizes that, fromthe perspective of charges, MIRP is not at a disad-vantage relative to ORP. Finally, when low caseloadMIRP surgeons were compared to low caseload ORPsurgeons, a charge disadvantage was identified forMIRP. This implies that some MIRP benefits, suchas shorter hospital stay and a lower complicationrate,8–10 may potentially be cancelled by MIRPcharges. For example, the cost increase may be re-lated to disposable equipments and/or longer opera-tive time. This consideration is important when im-plementing a new MIRP program is contemplatedand staffing is provided by junior MIRP users. More-over, this finding implies that junior MIRP usersshould be tutored by more expert MIRP surgeons todecrease the duration of the learning curve of lowcaseload surgeons.

These findings indicate that MIRP charges aremore sensitive to the effect of caseload than ORPcharges. This phenomenon is supported by the widerdistribution of MIRP caseload tertile specific mediancharges vs those of ORP (range $28,780 to $41,765vs $32,726 to $35,642). This distribution supportsour hypothesis that greater MIRP complexity mayexert an important effect on charges. Moreover, fa-miliarity with ORP since the early 1990s may alsocontribute to these findings. Accordingly less expertsurgeons contributed to higher charges, and more sofor MIRP than for ORP.

To our knowledge there are no other populationbased comparisons between MIRP and ORP chargesafter stratifying for caseload. Prior reports relied onsubstantially smaller patients counts and focused onindividual institution data.11,12 For example, in thelargest study Bolenz et al assessed 643 patientstreated with ORP (161), LRP (220) and robot as-sisted RP (262).11 Despite its sample size disadvan-tage the study benefitted from several methodologi-cal advantages. For example, the group focused onthe itemized direct cost of the various components ofeach surgical technique. This type of detailed anal-ysis is more feasible and reliable for a single insti-

tution than for a population based database.

Previous studies showed a cost disadvantage forMIRP relative to ORP.11,12 However, the smallersample size in previous reports precluded investi-gators from examining their data according to case-load.11,12 Notably when caseload stratification wasnot done, the 2 surgery types in our study resultedin virtually the same median charges. Accordinglylarge sample size is necessary to address the effectof caseload on RP charges. It may be argued thatour study is the first that provides such analysis.

Despite its strengths the use of population baseddata sets has several limitations. 1) The lack ofpathological information did not allow us to adjustfor tumor characteristics. 2) Our data may includeMIRPs that were done with or without roboticassistance, ie LRP. The advent of robotics hasvirtually eliminated LRP in the United States.3

Thus, a small fraction of all procedures may rep-resent LRP. The mentioned limitations wereshared with several previous reports.6,19,20 3) Ourcharges cover the period from hospital admissionto discharge. Additional costs, such as the cost oflate complications, readmission or recurrencetreatment, were not captured in the current dataset. Moreover, our analysis does not provide de-tails on the charges of each subunit, such as oper-ation room, anesthesia and equipment charges.Also, we examined charges rather than actualcosts in our analysis. However analogous these 2variables may be, the actual amount of health caredollars consumed by MIRP and ORP surgeons re-mains unknown. 4) Our analysis did not adjust forhospital type, eg large academic institutions vsreferral hospital vs small scale community hospi-tals. Possibly such adjustment would have helpedexplain some variations in caseload and its rela-tionship with charges. Unfortunately this infor-mation was not available in the Florida database.5) The calculated caseload does not account forprevious surgical experience. This is shared byvirtually all previous studies addressing a similartopic.12,17,19 6) Our findings suggest the regional-ization of RP in order of preference in the hands ofhigh caseload MIRP surgeons first, then high case-load ORP surgeons, then intermediate caseloadMIRP or ORP surgeons and finally low caseloadORP surgeon. However, those recommendationscannot be invariably followed. The existing infra-structure may not allow an increase in the numberof RPs done by high caseload surgeons. Similarlyhigh caseload surgeons may not have the individ-ual capacity to operate on more patients. Finally,not all patients accept referral to high caseloadsurgeons due to personal, family and social condi-

tions.

IMPACT OF CASELOAD ON PROSTATECTOMY TOTAL HOSPITAL CHARGES 861

CONCLUSIONS

Overall median charges are virtually the same forMIRP and ORP. However, MIRP charges are moresensitive to the effect of caseload, as evidenced by the

wider distribution of median MIRP caseload specific

REFERENCES

facts or fiction? Lancet 2007; 369: 723. order-data.shtml. Accessed

charges relative to those of ORP. After adjusting forcovariates high caseload MIRP resulted in the lowestcharges relative to all other MIRP and ORP categories.These observations should be considered for global

and/or institutional budget considerations.

1. Cooperberg MR, Lubeck DP, Meng MV et al: Thechanging face of low-risk prostate cancer: trendsin clinical presentation and primary management.J Clin Oncol 2004; 22: 2141.

2. Heidenreich A, Aus G, Bolla M et al: EAU guide-lines on prostate cancer. Eur Urol 2008; 53: 68.

3. Murphy DG, Bjartell A, Ficarra V et al: Downsidesof robot-assisted laparoscopic radical prostatec-tomy: limitations and complications. Eur Urol2010; 57: 735.

4. Descazeaud A, Peyromaure M and Zerbib M: Willrobotic surgery become the gold standard forradical prostatectomy? Eur Urol 2007; 51: 9.

5. Box GN and Ahlering TE: Robotic radical prosta-tectomy: long-term outcomes. Curr Opin Urol2008; 18: 173.

6. Hu JC, Gu X, Lipsitz SR et al: Comparative effec-tiveness of minimally invasive vs open radicalprostatectomy. JAMA 2009; 302: 1557.

7. Rocco B and Djavan B: Robotic prostatectomy:

8. Ficarra V, Novara G, Artibani W et al: Retropubic,laparoscopic, and robot-assisted radical prosta-tectomy: a systematic review and cumulativeanalysis of comparative studies. Eur Urol 2009;55: 1037.

9. Guazzoni G, Cestari A, Naspro R et al: Intra- andperi-operative outcomes comparing radical retro-pubic and laparoscopic radical prostatectomy: re-sults from a prospective, randomised, single-sur-geon study. Eur Urol 2006; 50: 98.

10. Remzi M, Klingler HC, Tinzl MV et al: Morbidity oflaparoscopic extraperitoneal versus transperito-neal radical prostatectomy versus open retropu-bic radical prostatectomy. Eur Urol 2005; 48: 83.

11. Bolenz C, Gupta A, Hotze T et al: Cost comparisonof robotic, laparoscopic, and open radical prosta-tectomy for prostate cancer. Eur Urol 2010; 57:453.

12. Burgess SV, Atug F, Castle EP et al: Cost analysisof radical retropubic, perineal, and robotic pros-tatectomy. J Endourol 2006; 20: 827.

13. FloridaHealthFinder.gov. Available at http://www.floridahealthfinder.gov/Researchers/OrderData/

January 18, 2010.

14. Hysterectomy Hospitalizations 1993-2002. HealthOutcome Series. Tallahassee: Agency for HealthCare Administration 2004; p 5.

15. Romano PS, Roos LL and Jollis JG: Adapting aclinical comorbidity index for use with ICD-9-CMadministrative data: differing perspectives. J ClinEpidemiol 1993; 46: 1075.

16. Birkmeyer JD, Stukel TA, Siewers AE et al: Sur-geon volume and operative mortality in theUnited States. N Engl J Med 2003; 349: 2117.

17. Birkmeyer NJ, Dimick JB, Share D et al: Hospitalcomplication rates with bariatric surgery in Mich-igan. JAMA 2010; 304: 435.

18. Abdi H: Bonferroni and Sidak corrections for mul-tiple comparisons. In: Encyclopedia of Measure-ment and Statistics. Edited by NJ Salkind. Thou-sand Oaks: SAGE Publications 2007.

19. Hu JC, Wang Q, Pashos CL et al: Utilization andoutcomes of minimally invasive radical prostatec-tomy. J Clin Oncol 2008; 26: 2278.

20. Begg CB, Riedel ER, Bach PB et al: Variations inmorbidity after radical prostatectomy. N Engl

J Med 2002; 346: 1138.