Accepte

ReprintHenry FMI 4820

0039-60

� 2012

http://d

1090

Laparoscopic versus openadrenalectomy: Another look atoutcome using the Clavienclassification systemLaura I. Eichhorn-Wharry, MD,a Gary B. Talpos, MD,a and Ilan Rubinfeld, MD, MBA,a,b Detroit, MI

Background. A laparoscopic approach to adrenalectomy has become the procedure of choice for mostadrenal resections. We hypothesized that laparoscopic adrenalectomy is less likely to result in intensivecare unit (ICU) level complications or death than open adrenalectomy, despite baseline comorbidity mix.Methods. Using the National Surgical Quality Improvement Program (NSQIP) participant use files for2005–2009, all laparoscopic and open adrenalectomies were identified by current proceduralterminology. Adverse outcomes tracked in NSQIP were mapped to Clavien level based on need for ICUcare or death. Univariate and multivariate analysis were used to compare groups.Results. There were 1,980 laparoscopic and 592 open procedures. Clavien 4 and 5 complicationsoccurred in 45 (7.6%) of open and 35 (1.8%) of laparoscopic operations. The univariate odds ratioshowed a 4.6-fold greater likelihood that a patient would have an ICU level complication (P < .001),and 4.9 odds ratio of death (P < .001) if an open rather than laparoscopic operation was performed.Regression modeling showed persistence of the protective effect of laparoscopy after adjusting for comor-bidities with a multivariate odds ratio of 3.3 (P < .001).Conclusion. The laparoscopic approach to adrenalectomy has an independent protective effect on ICUlevel complications and mortality when compared with open procedures. This correlation persists aftercorrecting for multiple comorbidities. (Surgery 2012;152:1090-5.)

From the Department of Surgery,a and the Center for Health Services Research,b Henry Ford Health System,Detroit, MI

SINCE ITS INTRODUCTION IN 1992, laparoscopic adre-nalectomy (LA) has gained popularity as theprocedure of choice for noninvasive adrenalmasses.1-6 Indeed, the criteria to use LA havebeen widening to include pheochromocytoma,7

larger masses,8-10 and even oncologic resec-tions11,12 when principles of surgical oncology13

may be safely applied.Prior comparisons of laparoscopic and open

adrenalectomies have repeatedly shown that LAprovides patient benefit, including improvementsin duration of stay, operative time, morbidity, andmortality.1,3,4,6,7,12,14-16 These studies typically havebeen retrospective or cohort studies and have beenlimited by the small volume that even busy centershave been able to acquire.

d for publication August 15, 2012.

requests: Ilan Rubinfeld, MD, Department of Surgery,ord Hospital, 2799 West Grand Boulevard, Detroit,2. E-mail: irubinf1@hfhs.org.

60/$ - see front matter

Mosby, Inc. All rights reserved.

x.doi.org/10.1016/j.surg.2012.08.020

SURGERY

Outcome measurement has become increas-ingly important in all specialties. Increasingly,hospital and physician quality is being tied toevidence-based, acuity-adjusted outcome mea-sures, such as postoperative complications incardiac surgery patients or postoperative readmis-sions. As this trend increases, insurance companiesare also experimenting with ‘‘pay-for-perfor-mance’’ models of reimbursement. In some cases,contracts are being directed to those with highvolumes or a lesser incidence of complications fora given procedure.

As indications for laparoscopic resectionbroaden, patient selection criteria widen to in-clude older patients and those with greater base-line medical illness. Advanced age17 and increasedcomorbidities18 are recognized as factors influenc-ing the success of a particular operation. With thenational movement toward measuring acuity andrisk-adjusted outcomes, surgeons need an under-standing of the best way to care for our patientsbased on the available evidence to reduce morbid-ity and avoid mortality in the context of optimizedrisk models.

SurgeryVolume 152, Number 6

Eichhorn-Wharry, Talpos, and Rubinfeld 1091

One way to classify these complications is withthe Clavien classification system, which provides avalidated, literature-based approach for categoriz-ing surgical complications. Clavien first reportedthe approach in 1992, with the goals of increasinguniformity of reporting, providing a means forcomparison between time periods and betweeninstitutions, to compare operative versus nonoper-ative measures, to adequately perform meta-analysis, and to identify and develop preoperativerisk factors and prognostic scores.19 The scoringsystem was modified in 2004 to better classify com-plications that were life threatening and perma-nently disabling; the system has been judged tobe simple, reproducible, comprehensive, andadequate.20,21

For our study, we determined that complicationsrequiring admission to the intensive care unit (ICU)or resulting in patient mortality would provide agood focus for our analysis, because they are oftenthe most expensive in terms of actual dollars spentand have the greatest impact on patient quality oflife. In an effort to evaluate a large volume ofadrenalectomies for the most severe complications,those resulting in ICU intervention and death, weused the American College of Surgeons (ACS)National Surgical Quality Improvement Program(NSQIP), which provides validated data on patientoutcomes for surgical patients. We sought to iden-tify whether the approach to adrenalectomy af-fected clinical outcomes of patients with the mostsevere complications.

METHODS

Data source and integrity. According to the datause agreement of the ACS, the 2005–2009 NSQIPparticipant use files were queried. The ACS NSQIPgathers data from multiple hospitals of diverse sizeand location throughout the United States. Dataare sampled in a validated manner within eachhospital, which provides a high degree of reliabilitybetween institutions. The participant use files datais compliant with the Health Insurance Portabilityand Accountability Act of 1996 and not identifiableto specific institution or surgeon. Preoperativevariables affecting outcomes as well as 30-daymorbidity and mortality are recorded.

Case selection. The American Medical Associa-tion current procedural terminology (CPT) codesfor LA (CPT 60650) and open adrenalectomy (OA;CPT 60540, 60545) were used to query the partic-ipant use files only for initial operation. The needfor conversion to open was not recorded.

We used a previously described mapping pro-cess to link NSQIP occurrences and Clavien

complications. The Clavien classification (TableI) was used to identify those complications thatwould likely lead to ICU admission (Clavien grade4). These were classified as septic shock, Q-waveinfarct, cardiac arrest, initiation of hemodialysis,unplanned reintubation, and prolonged ventila-tory requirements. Mortalities are classified as Clav-ien grade 5.

The recorded discharge International Classifi-cation of Diseases (ICD-9) codes were thenmapped to the following categories, ‘‘benign non-functional,’’ ‘‘functional,’’ and ‘‘malignancy.’’ ICD-9 codes for adrenal mass with uncertain behaviorwere mapped to the benign group. ICD-9 cate-gories that were incomplete or that did not fit intoone of these divisions were mapped to an ‘‘other’’category.

Multivariate logistic regression analysis was thenperformed to control for those variables shown toaffect outcome in prior studies and for surgeonselection bias. American Society of Anesthesiolo-gists (ASA) class, wound classification, preopera-tive functional status, preoperative albumin,emergent or not, and preoperative admissionwere used in our analysis. In addition, malignancywas compared with the other tumor types and tothe open approach.

Statistical analysis. Statistics were calculated us-ing SPSS. Chi-square analysis was used. P < .05 wasconsidered significant. Multivariate logistic regres-sion analysis was used to adjust for ASA class,wound classification, preoperative functional sta-tus, preoperative albumin, emergent or not, preop-erative admission, and tumor type. Continuousdata are presented as mean values ± standard devi-ations. This study was approved by our institutionalreview board.

RESULTS

Of the 2,572 adrenalectomies performed be-tween 2005 and 2009, 1,980 (77%) were LA and592 (23%) were OA procedures. The majority(98.4%) of procedures were performed by generalsurgeons with the remainder by urologists (1.4%),vascular surgeons (0.3%), or gynecologists(<0.1%).

Demographics were similar for each group(Table II), with a mean age of 53.1 ± 13.8 and53.3 ± 15.2 years for LA and OA, respectively. Over-all, 61% of patients were female, with 60% femalein the LA group and 62% female in the OA group.No significant differences were found in the cate-gories of race, ASA classification, or preoperativealbumin levels. The open group had a greater pro-portion of malignancy and the laparoscopic group

Table I. Classification of surgical complications

Grade Definition

1 Any deviation from postoperative coursewithout need for pharmacologicaltreatment or surgical, endoscopic, orradiologic intervention. Allowedtherapeutic regimens are drugs such asanti-emetics, antipyretics, analgesics,diuretics, electrolytes, andphysiotherapy. Includes wounds openedat bedside.

2 Requires pharmacologic treatment withdrugs other than those allowed forgrade 1 complications. Bloodtransfusions and parenteral nutritionincluded.

3 Requires surgical, endoscopic, orradiologic intervention

3a Intervention with patient not undergeneral anesthesia

3b Intervention with patient under generalanesthesia

4 Life-threatening complications (includingCNS complications) requiring IC/ICUmanagement

4a Single organ dysfunction (includesdialysis)

4b Multi-organ dysfunction5 DeathSuffix ‘‘d’’ If patient is experiencing complication at

the time of discharge, the suffix ‘‘d’’ fordisability is added to respective grade ofcomplication. This indicates need forfollow-up visits for full evaluation of thecomplication.

CNS, Central nervous system; IC/ICU, intensive care unit.

SurgeryDecember 2012

1092 Eichhorn-Wharry, Talpos, and Rubinfeld

had a greater proportion of benign nonfunctionaltumors and functional tumors. Mean duration ofstay was 2.8 ± 4.6 days for LA and 6.73 ± 6.9 daysfor OA. Mean operative time was 146 ± 67 minutesfor LA and 186 ± 97 minutes for OA.

Combined risk of Clavien 4 and 5 complicationswas 4.572 with a 95% confidence interval of 2.910–7.183 when the open approach was used. Similarly,the odds ratio of mortality was 4.9 with a confi-dence interval of 2.091–11.561. When assessed byspecific complication (Table III), all Clavien 4 and5 complications were significantly more likely inpatients undergoing OA, except Q-wave infarctand pulmonary embolism.

Multivariate regression analysis was performedfor each of the primary outcomes to account forASA class, wound class, preoperative functionalstatus, preoperative albumin level, emergency case,

and preoperative inpatient status. The probabilityof having any of the Clavien 4 level complications(Table IV) was 3.3 times greater for the openapproach than for the laparoscopic cohort. Themultivariate odds ratio for mortality was 4.5 (P <.001), even after correcting for the aforemen-tioned comorbidities.

We then compared malignancy status with theopen approach using logistic regression (Table V).The odds ratio for Clavien 4 or 5 complications was2.69 with 95% confidence interval of 1.6–4.3(P < .001).

DISCUSSION

Prior studies have shown clearly that LA is ofbenefit to patients from a morbidity and mortalitystandpoint. Our work affirms that conclusion. TheClavien classification system provides a frameworkthat can be used to further identify multiple levelsof complications to better understand the out-comes for our patients.

Interestingly, OA was found to be significantlymore morbid for all Clavien 4 and 5 complications,except for pulmonary embolism, where only atrend toward superiority was identified. We werenot able to elucidate the reasons for this. The rateof pulmonary embolism remains less in LA thanfor OA and may become significant as more yearsof data are evaluated.

Although the indications for laparoscopy havebeen increasing, the 2 methods are still compared,because no randomized trial comparing the 2methodologies has been performed, nor is it likelyto be performed given the body of data thatoverwhelmingly supports the use of laparoscopy.This finding creates some controversy regardingthe best approach with respect to early or confinedmalignancy. What remains unclear is whether thebenefits from a specific approach are significantenough to warrant further investigation or tosupplant the currently advocated OA for malig-nant indications. Clearly, for benign or functionalneoplasms, the benefits of laparoscopy suggest thatthe approach is the favored method.

There are limitations to our study. The ACS-NSQIP is a database designed to identifyinstitution-level outcomes data; it was not designedto capture data specific to any individual proce-dure or diagnosis. It is compliant with the HealthInsurance Portability and Accountability Act of1996, and does not identify surgeon or institutionspecific variables, which may affect outcome. Ad-ditionally, as a product of the ACS, the amount ofspecialty-specific data, such as urologic data, arenot included. The database is also general and

Table II. Patient demographics

Laparoscopic (n = 1,980) Open (n = 592) P value

Age (yrs) 53.1 ± 13.8 53.3 ± 15.2 NSGender, n (%) NS

Female 1,223 (60.0) 355 (61.8)Male 757 (38.2) 237 (40)

Race, n (%) NSCaucasian 1,519 (76.7) 442 (74.6)Black 212 (10.7) 71 (12)Asian 33 (1.6) 8 (1.6)Hispanic 26 (1.3) 6 (1.0)Other/unknown 190 (9.5) 62 (10.6)

ASA class, n (%) NS1 33 (1.7) 12 (2.0)2 753 (38.0) 183 (30.9)3 1,090 (55.1) 344 (58.1)4 102 (5.2) 49 (8.3)5 0 (0.0) 4 (0.7)

Functional status, n (%) <.001Independent 1,941 (98) 560 (94.6)Partially dependent 32 (1.6) 23 (3.9)Totally dependent 7 (0.4) 9 (1.5)

Comorbidities, n (%)CADPrior PCI/CABG 143 (7.2) 47 (7.9) NSMI in past 6 mos 18 (0.9) 11 (1.9) .05CHF past 30 days 19 (1.0) 10 (1.7) NSAngina past 30 days 13 (0.7) 2 (0.3) NS

DiabetesIDDM 125 (6.3) 44 (7.4) NSNIDDM 234 (11.8) 78 (13.2) NSCOPD 92 (4.6) 30 (5.1) NSHTN 1,420 (71.7) 377 (63.7) <.001

Preoperative albumin 4.0 ± 0.54 3.9 ± 0.64 NS

ASA, American Society of Anesthesiologists; CABG, coronary artery bypass grafting; CAD, coronary artery disease; CHF, congestive heart failure; COPD,chronic obstructive pulmonary disease; HTN, hypertension; IDDM, insulin-dependent diabetes mellitus; MI, myocardial infarction; NIDDM, non–insu-lin-dependent diabetes mellitus; PCI, percutaneous coronary intervention.

Table III. Incidence of postoperative complications and univariate odds ratio of occurrence

Laparoscopic(n = 1,980), n (%)

Open (n = 492),n (%) Odds ratio

95% confidenceinterval P value

Q wave infarct — —Cardiac arrest 1 (0) 3 (0.1) 10.08 1.047–97.084 .013Pulmonary embolism 10 (0.5) 7 (1.2) 2.357 0.893–6.220 .074Reintubation 14 (0.5) 21 (3.5) 5.165 2.610–10.221 <.001Prolonged ventilation 14 (0.7) 25 (4.2) 6.192 3.713–77.794 <.001Septic shock 9 (0.5) 18 (3.0) 6.868 3.069–15.369 <.001Dialysis 2 (0.1) 10 (1.7) 16.993 3.713–77.794 <.001Mortality 9 (0.5) 13 (2.2) 4.917 2.091–11.561 <.001Any class 4 or 5 complication 35 (1.8) 45 (7.6) 4.572 2.91–7.183 <.001

SurgeryVolume 152, Number 6

Eichhorn-Wharry, Talpos, and Rubinfeld 1093

does not include detailed data specific to adrenal-ectomy, such as need for conversion to open orhypertensive crisis during pheochromocytoma

resection, which would be useful to identify. Nordoes it include the inability to determine histologicdiagnosis or grading for the oncologic diagnosis.

Table IV. Risk-adjusted odds ratio ofcomplications

OddsRatio

95% confidenceinterval P value

Pulmonary embolism 0.38 0–9.137 .242Reintubation 2.917 0.924–9.21 .068Prolonged Ventilation 3.18 0.967–10.45 .057Septic shock 3.987 1.032–15.406 .045Dialysis 15.68 1.299–189.5 .030Mortality 5.536 1.357–22.58 .017Any class 4 or 5

complication3.301 1.020–5.689 <.001

Table V. Results

Laparoscopic(n = 1,980)

Open(n = 592) P value

Mean OR time (min) 146 ± 67 186 ± 97 .001Mean duration of stay

(days)2.8 ± 4.6 6.7 ± 6.9 .001

Discharge diagnosis,n (%)

.001

Nonfunctionalneoplasm

1,193 (60.3) 259 (43.8)

Functional neoplasm 489 (24.7) 109 (18.4)Malignancy 82 (4.1) 112 (18.9)Other 216 (10.9) 112 (18.9)

OR, Operating room.

SurgeryDecember 2012

1094 Eichhorn-Wharry, Talpos, and Rubinfeld

Postoperative adrenal insufficiency was notrecorded. Furthermore, size was not included inthe database and may play a role in the decision toperform a laparoscopic or open case. The currentCPT system does not account for the differences intransabdominal or retroperitoneal approach, northe use of surgical robotics to perform adrenalec-tomy. It is unclear how the use of robotics willchange outcomes. This cannot be elucidated fromany data set relying on current CPT coding.

In conclusion, using the NSQIP database toclassify complications of adrenalectomy with theClavien system allows us an opportunity to criticallyappraise the role surgical approach plays on short-term morbidity and mortality. Our data suggestresection of the adrenal by laparoscopic methodprovides improved outcomes related to the mostserious complications and death in the near term.The effect persists when multivariate analysis cor-rects for differences in patient comorbidities andtumor type. It is unclear whether the gains made inthe near term would be negated by a potentiallyincomplete oncologic resection in the case ofadrenocortical cancer confined to the adrenal. By

continuing to critically appraise the data, we maybe able to determine whether the gains made inthe near term by approach warrant the risks ofusing the laparoscopic approach in this instance.For other indications, it is likely that laparoscopyshould be used when technically feasible. Perhapsusing a data consortium approach to trackoutcomes specific to adrenal resection would allowfor a more detailed analysis to illuminate thiscontroversy.

The authors thank Sarah Whitehouse for assistancewith manuscript preparation.

REFERENCES

1. Prinz RA. A comparison of laparoscopic and open adrenal-ectomies. Arch Surg 1995;130:489-92.

2. Saunders BD, Wainess RM, Dimick JB, Upchurch GR, Doh-erty GM, Gauger PG. Trends in utilization of adrenalectomyin the United States: have indications changed? World JSurg 2004;28:1169-75.

3. Murphy MM, Witkowski ER, Ng SC, et al. Trends in adre-nalectomy: a recent national review. Surg Endosc 2010;24:2518-26.

4. Lee J, El-Tamer M, Schifftner T, et al. Open and laparo-scopic adrenalectomy: analysis of the National SurgicalQuality Improvement Program. J Am Coll Surg 2008;206:953-9.

5. Kebebew E, Siperstein AE, Duh QY. Laparoscopic adrenal-ectomy: the optimal surgical approach. J LaparoendoscAdv Surg Tech A 2001;11:409-13.

6. Gupta PK, Natarajan B, Pallati PK, Gupta H, Sainath J, Fitz-gibbons RJ Jr. Outcomes after laparoscopic adrenalectomy.Surg Endosc 2011;25:784-94.

7. Mellon MJ, Sundaram CP. Laparoscopic adrenalectomy forpheochromocytoma versus other surgical indications. JSLS2008;12:380-4.

8. Zografos GN, Farfaras A, Vasiliadis G, et al. Laparoscopic re-section of large adrenal tumors. JSLS 2010;14:364-8.

9. Parnaby CN, Chong PS, Chisholm L, Farrow J, Connell JM,O’Dwyer PJ. The role of laparoscopic adrenalectomy foradrenal tumours of 6 cm or greater. Surg Endosc 2008;22:617-21.

10. Castillo OA, Vitagliano G, Secin FP, Kerkebe M, Arellano L.Laparoscopic adrenalectomy for adrenal masses: does sizematter? Urology 2008;71:1138-41.

11. Saunders BD, Doherty GM. Laparoscopic adrenalectomyfor malignant disease. Lancet Oncol 2004;5:718-26.

12. Brix D, Allolio B, Fenske W, et al. Laparoscopic versus openadrenalectomy for adrenocortical carcinoma: surgical andoncologic outcome in 152 patients. Eur Urol 2010;58:609-15.

13. Miller BS, Ammori JB, Gauger PG, Broome JT, HammerGD, Doherty GM. Laparoscopic resection is inappropriatein patients with known or suspected adrenocortical carci-noma. World J Surg 2010;34:1380-5.

14. Ramachandran MS, Reid JA, Dolan SJ, Farling PA, RussellCF. Laparoscopic adrenalectomy versus open adrenalec-tomy: results from a retrospective comparative study. UlsterMed J 2006;75:126-8.

15. Porpiglia F, Fiori C, Daffara F, et al. Retrospective evaluationof the outcome of open versus laparoscopic adrenalectomyfor stage I and II adrenocortical cancer. Eur Urol 2010;57:873-8.

SurgeryVolume 152, Number 6

Eichhorn-Wharry, Talpos, and Rubinfeld 1095

16. Park HS, Roman SA, Sosa JA. Outcomes from 3144 adrenal-ectomies in the United States: which matters more, surgeonvolume or specialty? Arch Surg 2009;144:1060-7.

17. Kazaure HS, Roman SA, Sosa JA. Adrenalectomy in olderAmericans has increased morbidity and mortality: an analy-sis of 6,416 patients. Ann Surg Oncol 2011;18:2714-21.

18. Kazaure HS, Roman SA, Sosa JA. Obesity is a predictor ofmorbidity in 1,629 patients who underwent adrenalectomy.World J Surg 2011;35:1287-95.

19. Clavien PA, Sanabria JR, Strasberg SM. Proposed classifica-tion of complications of surgery with examples of utilityin cholecystectomy. Surgery 1992;111:518-26.

20. Dindo D, Demartines N, Clavien PA. Classification of surgi-cal complications: a new proposal with evaluation in a co-hort of 6336 patients and results of a survey. Ann Surg2004;240:205-13.

21. Clavien PA, Barkun J, de Oliveira ML, et al. The Clavien-Dindo classification of surgical complications: five-year ex-perience. Ann Surg 2009;250:187-96.

DISCUSSION

Dr Quan-Yang Duh (San Francisco, CA): It’s nice thatyou are using the NSQIP data to look at this. My mainconcern with this paper is what you mentioned in termsof the limitations. And probably the way to get away fromthat is to somehow stratify, as you tried to do by diagno-sis. Even though I’m obviously partial to the laparoscopicinstead of an open approach, I think it would be moreconvincing if you were to show, for example, in thatgroup of gastrinoma patients, that an open versus lapa-roscopic, that there is a significant difference in bothICU use and mortality.

Because if you just look at most of the people herethat do adrenalectomy, my patients that I do open, I canpromise you, it’s a whole lot worse than my patients thatI do laparoscopically. So comparing them would not tellme that laparoscopic is any better.

Dr Laura Eichhorn-Wharry (Detroit, MI): We can, butthe numbers are very small by diagnosis. So the functionaltumor group, I think, only had 400 patients in it. And thento divide that by pheochromocytoma, aldosteronoma,and other functional tumors, I think we would losesome power. As the data continue to be accrued overtime, I think a subgroup analysis would be possible.

DrMichael Yeh (Los Angeles, CA): I’m concerned thatthe potential for confounders is formidable here. And I’munsure if your statistical methods are sufficiently rigorousto allow you to support your conclusions. There are a mil-lion reasons, as you and Dr Duh have alluded to, thatwould cause a surgeon to select a certain type of approach.These include some of the things you’ve suggested, but

also basics things, like what about the availability of sur-geon, geographic factors, where does the patients live inrelation to the nearest tertiary center?

There are 2 statistically valid ways of dealing with thistype of bias. And I would call your attention to the currentissue of JAMA, which talks about instrumental variables aswell as propensity scoring. And I would encourage you touse one of these techniques. In particular, there’s a paperby Freischlag and associates, comparing open versus en-dovascular AAA, which uses propensity scoring.

Dr Christopher R. McHenry (Cleveland, OH): Mycomment would be very similar to Dr Duh and Dr Yeh,but it’s really, were your laparoscopic and open adrenal-ectomy groups really comparable in terms of tumor sizeand pathology?

It’s conceivable that the disproportionate number ofpatients, for instance, that had a large adrenocorticalcancer underwent open adrenalectomy. And as a result,you would expect that morbidity and mortality would behigher in that group. So I think that’s a really importantissue with your data.

Dr Laura Eichhorn-Wharry: It is. What, I think, is a lit-tle concerning is, out of the open group, nearly 60%were benign or nonfunctional tumors. So there mustbe a reason why these were done open. But is the reasonwhy a laparoscopic approach wasn’t attempted, werethey converted? It’s unclear from our data.

DrRichardHodin (Boston,MA): I share, I think, every-one’s concern about the sort of bias here. And 1 question,I’m assuming you can’t get tumor size from this data, be-cause that probably explains a lot of the decision making.

But I wonder if also, using the NSQIP database theway you are, whether you can get data within a singleinstitution, and that’s sort of getting at Dr Duh’s point.

Dr Laura Eichhorn-Wharry: It’s not identifiablewithin a single institution.

Dr Richard Hodin: I’m not saying you have to identifywhat the institution is, but can you separate the data outfrom institution to know that a certain number of pa-tients came from a single institution. And then, withinthat group, open versus laparoscopic?

Dr W. Barry Inabnet (New York, NY): The NSQIP dataregistry is a wonderful thing, of course, but your presen-tation highlights some of the weaknesses of NSQIP whenyou are dealing with a specific disease system. And thisunderscores the importance of developing a disease-specific registry for adrenal surgery, for example, be-cause then we could tease out a lot of these detailsthat are so important to adrenal surgeons worldwide.