The Effect of Surgical Volume and the Provision of Residency and Fellowship Training on Complications of Major Hepatic Resection
- First Online:
- Cite this article as:
- Kohn, G.P. & Nikfarjam, M. J Gastrointest Surg (2010) 14: 1981. doi:10.1007/s11605-010-1310-z
- 94 Views
Positive volume–outcomes relationships have been demonstrated for hepatic resection using arbitrary criteria to define high-volume centers. The safety of training programs has not been evaluated. The association of surgical volume, as a continuous variable and the influence of a surgical residency and a fellowship program on outcomes after major hepatectomy were determined.
The Nationwide Inpatient Sample (NIS) was queried from 1998 to 2006. Quantification of patients’ comorbidities was made using the Charlson index, and mortality, and complication rates were determined. Institutions’ annual case volumes were correlated with risk-adjusted outcomes over time, as well as presence or absence of residency or fellowship training program using logistic regression modeling.
A total of 5,298 major hepatectomies were recorded, representing a weighted nationwide total of 26,396 cases. In-hospital unadjusted mortality for the study period was 6%. Adjusting for comorbidities, greater major hepatectomy volume was associated with improvements in the incidence of most measured complications, with plateauing of mortality of between 2% and 3% at approximately 50 cases per year. The mortality rate increased once greater than approximately 70 cases were performed per annum. Hospitals supporting a surgical residency program had lower overall morbidity and mortality. A fellowship program however was not associated with overall lower morbidity and mortality and appeared to result in a higher rate of certain complications.
Greater annual major hepatectomy volume improves outcomes with reduced mortality up to a certain point. The presence of surgical residency program but not a fellowship program is associated with reduced predicted morbidity and mortality.
KeywordsMajor hepatectomyFellowshipResidencyMortalityMorbiditySurgical volumeVolume-outcomeCharlson index
Hepatic resection has previously been associated with high morbidity and mortality.1–3 Improvements in technique and peri-operative care have resulted in significant reductions in morbidity, with reports of zero mortality in some centers.4,5 The overall morbidity and mortality associated with hepatic resection however appears to be under-reported, based on large database reviews.6 Identification of potentially modifiable factors associated with morbidity and mortality following major liver resection is therefore important.
A number of studies identify surgical volume as an important determinant of postoperative mortality following advanced surgical procedures.7–10 In cases of major hepatic resection, there is a clear association between mortality and surgical volume when arbitrary cut-offs are used to differentiate high- from a low-volume centers.6,11,12 There is no consistency or clear reasoning for using certain cut-off thresholds with definitions ranging from 15 to 50 cases per annum used to define a high-volume center.6,11,12 No previous population studies have examined the association of surgical volume as a continuous variable with inpatient mortality following major hepatic resection.
In addition to surgical volume, it is conceivable that surgical centers offering a residency program or a fellowship in hepatobiliary surgery may also be associated with lower morbidity and mortality. As far as is known, this has not been previously examined. It could be hypothesized that academic centers with a surgical residency or fellowship program provide more focused specialization, which could result in better outcomes, independent of surgical volume and patient comorbidities.13
The aim of our study was to analyze a large inpatient database to determine the association of surgical volume, as a continuous variable, with mortality following major hepatectomy. Secondarily, the outcomes of patients following major hepatectomy in centers offering an Accreditation Council for Graduate Medical Education (ACGME)-accredited surgical residency program or a fellowship in hepatobiliary surgery were compared with centers that offered no such programs, controlling for surgical volume and patient comorbidities.
National Inpatient Study Database
National Inpatient Study (NIS) database covering the years 1998–2006 was queried. This is the largest all-payer inpatient care databases in the USA, containing data from approximately eight million hospital stays each year. The latest release is the 2006 database contains all discharge data from 1,045 hospitals located in 38 States, approximating representing a 20% stratified sample of all non-Federal, short-term, general, and other specialty hospitals in the USA.14
Creation of Liver Resection Dataset
A dataset was created by merging the core and hospital files and so that only patients having undergone what was considered a major hepatectomy were included in this study. This was based on ICD-9-CM code 50.3, which defines hepatic lobectomy or greater. Patients undergoing minor hepatectomy or wedge resection (ICD-9 codes 50.2, 50.22, 50.29), and liver transplantation (ICD-9 codes 50.4 and 50.5) were excluded from analysis. Liver operations in patients whose stay in hospital was less than or equal to 23 h are not captured by the NIS database. This compromises mainly minor liver biopsy or minor resectional procedures. Pediatric patients less than or equal to 17 years of age were excluded. To calculate nationwide case volumes, the NIS-supplied discharge-level weight was applied to calculations. At all other times, the unweighted NIS cohort was utilized for calculating standard errors and performing regression analyses.
Identification of Residency and Fellowship Programs
Information regarding the presence of a fellowship program in each year of the study period was taken from the following: (a) The Fellowship Council’s (FC) webpage (a total of 89 institutions submitted data in 2006), b) Society of Surgical Oncology (SSO) webpage (11 institutions submitted data) and (c) The International Hepato-Pancreato-Biliary Association (IHPBA) Fellowship listings (ten institutions submitted data).15–17 The exact years of existence of the fellowship program could be determined for Fellowships listed on the Fellowship Council’s webpage. Some programs were listed on both IHPBA and FC websites. In the case of IHPBA and SSO listed programs, the presence of listing on the website was assumed to indicate the presence of such a fellowship throughout all the years of the study. It was also assumed that a Fellowship when listed was actually filled by a fellowship candidate in the year of listing. A teaching hospital was defined within the NIS as a hospital with residents in any specialty and meeting any of the following criteria: Accreditation Council for Graduate Medical Education (ACGME) residency training approval (in any specialty), membership in the Council of Teaching Hospitals, or a ratio of full-time equivalent interns and residents to beds of 0.25 or higher. Hospitals defined as having an ACGME-accredited general surgical residency were defined as a separate group. Details of such a surgical residency program were obtained by combining information from the American Medical Association’s FREIDA database and the listings of accredited programs on the ACGME webpage.18,19
Identification of Patient Comorbidity
Comorbidity scores were applied to each inpatient stay record, using the Deyo adaptation of the Charlson comorbidity index.20,21 This validated index allocates a score between 0 and 35, with a higher score indicating more comorbidity. The comorbidities examined include: myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, pulmonary disease, connective tissue disease, peptic ulcers, chronic liver disease, hemiplegia, renal disease, diabetes, malignancy, leukemia, metastatic cancer, and acquired immune deficiency syndrome.
Defining Mortality and Morbidity
Peri-operative complications were added based on ICD-9-CM codes, in a similar manner to that described by Santry.20 The diagnosis of “any complication” was made if the patient “died during hospitalization” field = 1, or if any of the NIS’s 15 diagnosis fields contained one of the following complications or procedure codes: abdominal drainage procedure (5,491), acute cerebrovascular accident (43100-43191, 4330-4339, 4340-43491), acute dialysis (3,895), acute deep venous thrombosis (4,538, 4,539), acute myocardial infarction (4,100–4,109), acute pulmonary embolism (4,151, 41,511, 41,519), acute renal failure (5,841–5,849), acute respiratory failure (51,881), adhesiolysis (5,451, 5,459), anastomotic leak (9,986), bacterial pneumonia (481, 485, 486, 4,820–4,829), cardiac complications (9,971), central nervous system complications (99,701–99,703), dialysis catheter insertion (3,995), foreign body removal (5,492), intraoperative hemorrhage (99,811), laparotomy (5,412), mechanical ventilation (967, 9,671, 9,672, 9,673), postoperative shock (9,980), reclosure of abdomen (5,461), respiratory tract complications (99,973), small bowel obstruction (5,600–5,609), splenectomy (4,143, 415), splenic injury (8,650–8,651), tracheostomy (311, 3,129), transfusion (9,904, 9,909), urinary complications (9,975), wound dehiscence (9,983, 99,831, 99,832), wound infection (9,985, 99,851, 99,859), wound seroma (99,813).
SAS 9.2 (SAS Institute, Cary, NC) was used to analyze the data. Logistic regression modeling was performed using Generalized Estimating Equations and assuming a binomial distribution of the data. This allowed certain covariables to be controlled for; these included annual improvements in outcomes and Charlson comorbidity index scores. Repeated measure analysis was performed with the experimental unit being hospital identification number clusters. We used logistic regression modeling to model the dichotomous response variables. We used compound symmetric correlation to account for expected correlation within individual hospitals. We used reference cell coding for our parameterization. The model was fit and empirical standard error estimates were generated which in turn were used to generate p values. A p value <0.05 was considered significant. Subsequently, the estimates were exponentiated to calculate an odds ratio and 95% confidence intervals.
We fit the models first with the quadratic terms, for which p values were generated. If statistical significance was determined, then the quadratic term was retained in the model. Otherwise, the quadratic was removed and a linear model was utilized. In interest of simplicity of presentation of our data, tables were presented using the linear model, while graphical representation of the case volume/outcome relationship allowed for demonstration of the model with the quadratic terms, which had reached significance.
Numbers of Major Hepatectomy
Unweighted number of major hepatectomy cases performed at various locations (% total hepatectomies)
All teaching hospitals
ACGME surgical training
Charlson Comorbidity Score and Unadjusted Mortality Rates
The Effect of Case Volume on Morbidity and Mortality
The incremental effect of each major hepatectomy on annual outcomes, controlling for year and for Charlson comorbidity scores
Odds ratio (95% CI)
Odds ratio (95% CI)
0.992 (0.987, 0.996)
Respiratory tract comps
0.993 (0.986, 1.000)
0.975 (0.967, 0.983)
Acute renal failure
0.989 (0.983, 0.995)
0.999 (0.991, 1.007)
0.994 (0.984, 1.004)
0.999 (0.993, 1.005)
0.986 (0.995, 0.999)
0.993 (0.985, 1.001)
0.983 (0.977, 0.995)
0.996 (0.989, 1.002)
0.992 (0.985, 0.999)
0.990 (0.982, 0.998)
1.000 (0.990, 1.011)
1.001 (0.996, 1.006)
1.007 (0.999, 1.014)
0.978 (0.958, 0.998)
Need to reclose abdomen
0.994 (0.983, 1.006)
0.989 (0.986, 0.993)
0.992 (0.986, 0.998)
1.000 (0.997, 1.004)
0.996 (0.991, 1.001)
0.986 (0.977, 0.996)
Nearly all analyzed complication categories trended towards an inverse correlation with volume, with mortality rate and rates of any complication, myocardial infarction, intraoperative hemorrhage, postoperative shock, splenectomy, bacterial pneumonia, respiratory failure, tracheostomy achieving statistically significant improvement. No complication was positively correlated with increasing case volume.
Relation of Surgical Residency on Morbidity and Mortality
The effect of surgical residency on major hepatectomy outcomes, controlling for year and for Charlson comorbidity scores
Odds ratio (95% CI)
Odds ratio (95% CI)
Respiratory tract comps
0.829 (0.683, 1.006)
0.815 (0.706, 0.941)
Acute renal failure
0.917 (0.786, 1.070)
1.114 (0.860, 1.443)
1.086 (0.755, 1.562)
1.088 (0.895, 1.323)
1.136 (0.861, 1.500)
1.037 (0.796, 1.351)
0.972 (0.756, 1.250)
0.886 (0.687, 1.143)
1.330 (0.993, 1.780)
1.029 (0.681, 1.554)
Need to reclose abdomen
1.046 (0.890, 1.230)
1.013 (0.887, 1.157)
Relation of Fellowship Program on Morbidity and Mortality
The effect of fellowship program on major hepatectomy outcomes, controlling for year, Charlson comorbidity score, and volume
Odds ratio (95% CI)
Odds ratio (95% CI)
0.931 (0.786, 1.103)
Respiratory tract comps
0.878 (0.701, 1.099)
0.855 (0.712, 1.027)
Acute renal failure
0.957 (0.797, 1.148)
1.233 (0.884, 1.722)
1.167 (0.780, 1.747)
1.366 (1.112, 1.678)
0.947 (0.776, 1.155)
1.375 (1.037, 1.825)
0.841 (0.667, 1.062)
1.425 (1.074, 1.890)
1.268 (0.954, 1.684)
1.303 (0.983, 1.728)
0.977 (0.643, 1.486)
1.247 (1.003, 1.550)
1.289 (0.899, 1.847)
1.034 (0.596, 1.795)
Need to reclose abdomen
1.477 (0.937, 2.330)
1.007 (0.889, 1.14,)
1.090 (0.891, 1.335)
0.969 (0.825, 1.138)
1.433 (1.111, 1.848)
1.334 (1.049, 1.697)
Hepatic resection is effective in the management of various liver tumors. There however continues to be significant morbidity and mortality associated with major hepatic resection based on large populations studies,6,11,12 despite observations of decreased mortality following liver resection over time.6 Several studies advocate that advanced surgical operations should be performed in high-volume centers,7–10 with arbitrary cut-off volumes used to differentiate between high- and low-volume centers. The true extent to which case volume improves outcomes following major hepatectomy has not been previously defined, where case volume is considered as a continuous variable. In addition, the effect of surgical residency and a fellowship training program on outcomes is unknown.
Multiple studies have previously demonstrated reduced mortality associated with advanced surgical procedures in high case volume centers.7–10 The majority of these studies have utilized the NIS database for analysis. Although not all patients from all States are captured by this database, it is the largest all-payer inpatient database in the USA.14 The mortality rate following liver resection in high-volume centers based on NIS data from 1998 to 2005 was significantly lower than in low-volume centers (2.6% versus 4.8%).12 The volume cut-off for a high-volume center was 20 or more cases. Volume cut-offs of 15 and 50 cases has also been previously examined.6,11 It is proposed that high-volume centers may provide greater overall specialization and care of patients undergoing complex operative procedures.8,22 The provision of specialized centers that attracts high case volume may be one factor accounting for these differences.
One must consider that mortality rates using the NIS database represent only in-hospital mortality, which is not necessarily reflective of 30–90 day postoperative mortality. It would be expected that the true 30 to 90 day postoperative mortality would be higher than determined from the NIS database. Patients in more specialized high case volume centers may be better streamlined for early home or rehabilitation discharge following surgery. If morbidity and mortality occurs after discharge, this would results in an apparent decrease in postoperative morbidity and mortality.
In our analysis of NIS data from 1998 to 2006, a small decrease in mortality was apparent over the study period. There was also clear reduction in morbidity and mortality following major hepatectomy with increasing case volume when controlling for year and for comorbidities using the Charlson comorbidity scoring system which has been previously validated and shown to be predictive of mortality.23–25 The greater the number of hepatectomies performed annually, the better the expected outcome, at least up to a point. It appears that at approximately 70 major hepatectomies per year, the mortality rate may rise. This is the first times such a U-shaped curve has been described in the field of hepatic surgery, though this mirrors findings previously reported in other surgical specialties.13 The argument that the high-volume centers may be tackling higher risk cases is partly rebutted by the aforementioned Charlson score-based risk adjustment data. As previously described, the Charlson score is a validated method of risk-stratification for surgical patients analysis with administrative databases. However, the authors acknowledge that it has not explicitly been validated in patients undergoing hepatic resection, and may therefore introduce some amount of systematic error into the analysis. It is postulated by the authors that at a certain volume, the facilities of the institution, both physical facilities and personnel, may be stretched enough to compromise patient outcomes. Alternatively, these facilities may deal with more advanced cases, requiring larger volumes of hepatic resection and higher risks of postoperative liver failure. These features cannot be determined from the NIS database. This clearly warrants further investigation, and has been undertaken by the authors as a future project. Regardless, it appears clear that case volume cannot be used as a perfect surrogate for surgical quality.
The association of a surgical residency or fellowship program, and morbidity and mortality following major hepatectomy has not been previously examined, although the teaching status of a hospital was not independently associated with mortality in a single reported study.6 That particular study examined both minor and major hepatectomy combined, but did not specifically examine accredited surgical residency programs and covered the period 1998–2004. According to our data assessing programs offering ACGME-accredited surgical residency, this was independently associated with overall reduced morbidity and mortality. This finding may be reflective of the high specialization of these units that are accredited for a surgical residency program, and has been similarly reported with regard to other operations such as esophagectomy.13 Given the advanced nature of major hepatectomy, these operative procedures are likely to be performed by attending surgeons than surgical residents. The overall influence of surgical residency on the technical aspects of major hepatectomy operations is therefore likely to be minimal.
The same findings were not apparent with programs offering surgical fellowships as identified by the Society of Surgical Oncology, the Fellowship Council, and the IHPBA listings.15–17 A Fellowship program was independently associated with an increase in specific complications following major hepatectomy. Our multivariate analysis took into consideration that many programs provide both surgical residency training and a Fellowship program. The specific factors that significantly increased were acute DVT, pulmonary emboli, cardiac complications (other than myocardial infarction), tracheostomy requirement, and wound dehiscence. Interestingly, there was no significant change in mortality or overall morbidity in centers offering a fellowship program. The exact reason for these finding is unknown, is likely to be multifactorial, and may be further evidence of a previous finding that in specialized centers earlier detection and commencement of management of complications lessens the effect of adverse events.26
It should be clearly noted that the performance of a hepatectomy in a hospital offering fellowship training does not guarantee that the fellow performed or was even a participant in the procedure. Furthermore, only a subset of Fellowship Council-accredited programs provide training to the Fellow in hepatic resection and similarly some hepatic surgery fellowships are provided outside the studies groups of Society of Surgical Oncology, the Fellowship Council, and IHPBA (for example, those programs accredited by the American Society of Transplant Surgeons). These limitations of the dataset will necessitate caution in the interpretation of associations between Fellowship training per se and outcomes, but will not nullify the association between the parent institution and the outcomes under consideration.
It should be noted that there was variability in the reporting of major liver resections by the various programs during certain time periods. It is possible that during some years, certain programs did not submit cases to the NIS or the NIS did not sample cases from those regions during specific periods.
Fellowship programs may perform more extensive liver resections such as extended lobectomy, which would be coded simply as a major liver resection. Such cases are more complex, take longer to complete and are at higher risk of complications. One may also hypothesize that increased morbidity may occur after advanced liver resections in which the fellow was the primary operator. Fellows may be more likely to be the primary operator for major hepatectomy than surgical residents and inexperience may produce more complications. Increased complications did not however translate to increased mortality that may be reflective of the high standard of postoperative care in such centers.
Socioeconomic status and insurance type was not corrected for in our study, which has previously been associated with mortality following liver resection.12 Also, our study included patients that had undergone at least a hepatic lobectomy, whereas others have included all patients undergoing hepatic resection.6 It is possible that not all patients undergoing major hepatectomy were identified based on ICD-9-CM classification as problems of both over- and under-capture of cases with ICD-9-CM-based searches of administrative datasets have previously been reported.27 Major hepatectomy is generally considered resection of three or more hepatic segments;28 this may not necessary involve a complete right or left hepatic lobectomy and some patients undergoing a major hepatectomy may have been coded as having a partial hepatectomy (50.22).
The predicted mortality following major hepatectomy decreases with increasing case volume without a specific volume cut-off, though a U-shaped curve exists with upturn at very high annual case volumes. Centers with surgical residency programs appear to be associated with reduced morbidity and mortality. Fellowship-associated programs have increased overall morbidity, without increased mortality. The factors related to these observed differences are only speculative and are worthy of further investigation.
For conception and design: GPK, MN; for acquisition of data: GPK; for analysis: GPK and MN; for drafting of manuscript: GPK, MN; for critical revision: GPK, MN; for statistical expertise: GPK. The authors acknowledge the general statistical assistance provided by JA Galanko, Ph.D. of the University of North Carolina at Chapel Hill, NC.