Understanding Mortality as a Quality Indicator After Esophagectomy Dustin M. Walters, MD, Timothy L. McMurry, PhD, James M. Isbell, MD, MSCI, George J. Stukenborg, PhD, and Benjamin D. Kozower, MD, MPH Departments of Surgery and Public Health Sciences, The University of Virginia, Charlottesville, Virginia

Background. Postoperative death is an important outcome after esophagectomy, and the Centers for Medicare & Medicaid Services currently uses 30-day mortality as a quality indicator for this operation. However, 30-day mortality may underestimate a patient’s true postoperative death risk. The purpose of this study was to evaluate different mortality definitions using a large registry of patients undergoing esophagectomy for cancer. Methods. Data were extracted from the Surveillance, Epidemiology and End Results-Medicare registry for patients with esophageal cancer who underwent esophagectomy between 2006 and 2009. Postoperative death was compared using four different definitions: 30-day, inhospital, perioperative (in-hospital or 30-day), and 90-day mortality. Hierarchical logistic regression models evaluated the association between patient and tumor characteristics with survival at 30 and 90 days and the ability of death to differentiate between good hospitals and those that perform poorly.

Results. We identified 634 patients from 188 hospitals. The 90-day mortality rate (13.3%) was more than double the 30-day mortality rate (6.0%) in this patient population. Advanced age and diagnosis of chronic obstructive pulmonary disease were associated with an increased risk of 90-day mortality. Good or poor performers could not be determined using the 30-day or 90-day mortality rate. Conclusions. There are clinically meaningful differences between postoperative mortality definitions after esophagectomy. Thirty-day mortality significantly underestimates a patient’s true risk of death because this number more than doubles at 90 days in this elderly, Medicare population. Although neither 90-day nor 30-day mortality are adequate quality measures after esophagectomy, 90-day mortality is a better outcome measure because it provides a better understanding of true death risk for the surgeon and patient.

E

Services (CMS), which derives its quality indicators from the Agency for Healthcare Research and Quality (AHRQ), collects and reports 30-day mortality data. Although 30-day mortality may be a reasonable quality measure in the initial postoperative period, it may also significantly underestimate the true risk of death after certain high-risk surgical procedures, including esophagectomy. The two principal reasons for this are recent improvements in critical care capabilities [5] and a lack of appreciation of the duration of postoperative sequelae, particularly for patients with postoperative complications. Accordingly, some authors have suggested that 90-day rather than 30-day mortality may more accurately predict postoperative risk for high-risk operations such as hepatectomy, colectomy, and pulmonary resection [6–9]. The primary aim of this study was to evaluate different definitions of postoperative mortality after esophagectomy: in-hospital, 30-day, perioperative (in hospital or 30-day), and 90-day mortality. For this purpose, we chose the Surveillance, Epidemiology, and End Results (SEER)-Medicare database, which captures a large percentage of elderly patients with cancer and thus may better represent national data than smaller, more specialized databases such as the STS-GTSD [10]. In addition, SEER-Medicare allows for collection of mortality data at 90 days, unlike other databases, such as the

sophageal cancer represents only 1% of all cancers in the United States, but with an estimated 17,990 new cases and 15,210 deaths in 2013 [1], it remains one of the deadliest cancers after diagnosis. Esophagectomy offers the best opportunity for cure and remains a cornerstone of therapy for patients with localized disease. Unfortunately, resection is coupled with high postoperative morbidity and mortality rates, despite advances in surgical technique and critical care. In current large series, morbidity and mortality are reported to be as high as 24% to 39% and 4% to 11%, respectively [2–4]. Mortality is frequently reported as death within 30 days of the operation. This is congruent with many administrative and national databases, including The Society of Thoracic Surgeons (STS)-General Thoracic Surgery Database (GTSD) and the American College of Surgeons National Surgical Quality Improvement Program (ACSNSQIP). Similarly, the Center for Medicare & Medicaid

Accepted for publication March 26, 2014. Presented at the Fiftieth Annual Meeting of The Society of Thoracic Surgeons, Orlando, FL, Jan 25–29, 2014. Address correspondence to Dr Kozower, University of Virginia, General Thoracic Surgery, Box 800679, Charlottesville, VA 22908-0679; e-mail: [email protected].

Ó 2014 by The Society of Thoracic Surgeons Published by Elsevier Inc

(Ann Thorac Surg 2014;-:-–-) Ó 2014 by The Society of Thoracic Surgeons

0003-4975/$36.00 http://dx.doi.org/10.1016/j.athoracsur.2014.03.041

2

WALTERS ET AL MORTALITY AFTER ESOPHAGECTOMY

STS-GTSD. We hypothesized that a definition of mortality within 90 days of the operation would capture a significantly higher number of deaths than other common definitions and might better estimate a patient’s true operative risk. Secondary aims included the determination of predictors for death after esophagectomy and appraisal of 30-day and 90-day mortality as a quality indicator by their ability to differentiate between goodhospitals and those that perform poorly.

Patients and Methods The University of Virginia Institutional Review Board for Health Sciences Research approved this study.

SEER-Medicare Database The SEER Program collects data on patients with cancer in the United States, including incidence, survival, demographics, tumor-specific data (eg, tumor site, morphology, and stage), and treatment. The National Cancer Institute, SEER Registry, and Centers for Medicare & Medicaid Services (CMS) have collaboratively linked the SEER Registry to Medicare claims data for eligible patients [11]. This linkage provides longitudinal information regarding survival, inpatient admissions, outpatient events, and health care claims for 93% of patients aged older than 65 years. The linked SEERMedicare database encompasses approximately a quarter of the population of patients with cancer and is broadly generalizable to the Medicare population.

Patient Population and Outcomes The 2006 to 2010 SEER Medicare database was queried to identify records of all patients older than 65 years with esophageal cancer who underwent esophagectomy. Exclusionary criteria included diagnosis at death or autopsy, records with nonmatching histology, patients who did not meet insurance criteria during the 3 months preceding esophagectomy, and those with less than 3 months of preceding Medicare data. The latter exclusion criterion was to ensure adequate preoperative comorbidity data. Demographic data collected included age, sex, race, and treating facility, and clinical data included year of operation, final pathologic stage, procedure type, and comorbidities. Comorbidities were collected using the SAS (SAS Institute, Cary, NC) search code provided by the National Cancer Institute based on inpatient files (MEDPAR), outpatient files (OUTSAF), and physician claims data (NCH) [12, 13]. The primary outcome was death from all causes among four different mortality definitions: 30-day, in-hospital, perioperative, and 90-day mortality. In-hospital mortality was defined as any death after the operation before the initial discharge from the hospital. Perioperative mortality was defined as 30-day or in-hospital mortality. All mortality data were derived from Medicare death certificate records within the SEER-Medicare database. Secondary outcomes included predictors of death at 30-days

Ann Thorac Surg 2014;-:-–-

and 90-days and validity of mortality as a quality measure.

Statistical Analyses To compare 30-day and 90-day mortality, we calculated the 95% confidence interval for the proportion of deaths occurring in the second and third months postoperatively. A hierarchical generalized logistic regression model was used to estimate 90-day mortality risk, with adjustments for data clustered by treatment provider. Model predictors were selected a priori from a literature review and frequency of occurrence within our data set. Modeling was performed with individual comorbidity variables. The contribution of each predictor of death to the overall predictive capacity of the models was assessed using the F test statistic. To test the utility of the mortality rate as a quality measure, we removed the hospital clustering effect to calculate an expected mortality rate for each hospital based on patient characteristics and compared this with each provider’s observed mortality rate. Bonferroni adjustment for multiple comparisons was also performed. All outcomes data were analyzed using SAS 9.3 software. Provider volume data is represented using R software [14] together with the ggplot package [15].

Results A total of 10,771 patients diagnosed with esophageal cancer in years 2006 to 2009 were identified in the SEERMedicare database. Patients were excluded who were diagnosed at death or autopsy (n ¼ 133), did not undergo esophagectomy (n ¼ 9,664), were younger than 65 years at diagnosis or had the operation within 3 months of turning 65 (n ¼ 116), had nonmatching histology (n ¼ 16), were enrolled in a Medicare health maintenance organization during the month of the operation or the 3 months prior (n ¼ 193), or had less than 3 months of preceding Medicare data in our data set at the time of the operation (n ¼ 15; Fig 1). We identified 634 patients who underwent esophagectomy for esophageal cancer and met the inclusion criteria. The Medicare providers performed a range of 1 to 31 cases, with more than half of providers performing only 1 esophagectomy per year on Medicare patients (Fig 2). Most patients were male and were between ages 65 and 75, with few patients aged older than 85 years (Table 1). Approximately a quarter of the patients were diagnosed in each calendar year 2006 through 2009. The pathology for a small number of patients was consistent with carcinoma in situ, and all stages of invasive cancer were well represented. Metastatic spread to lymph nodes (N1 disease) had occurred in approximately half of the patients. Nearly half received some sort of neoadjuvant therapy (chemotherapy, radiotherapy, or both), with 41.8% receiving neoadjuvant radiotherapy. The rate of comorbid disease was high. The most frequent comorbidities were chronic obstructive pulmonary disease (COPD) and diabetes, whereas peripheral arterial disease, heart failure, coronary artery disease, and renal failure were present at lesser frequencies (Table 1).

Ann Thorac Surg 2014;-:-–-

WALTERS ET AL MORTALITY AFTER ESOPHAGECTOMY

3

Fig 1. Flow chart demonstrates exclusion criteria.

Patients diagnosed with esophageal cancer (n=10,771)

Patients diagnosed at death/autopsy (n=133)

Surgery not performed (n=9,664)

Non-matching histology (n=16)

Age < 65 years (n=116)

Less than 3 months of preceding Medicare data (n=15)

Did not meet insurance criteria 3 months prior to surgery (n=193)

Patients meeting all inclusion criteria (n=634)

The mortality rate after esophagectomy varied greatly across the different definitions (Table 2). The 30-day mortality rate in this series was 6.0%. In-hospital mortality and perioperative mortality were slightly higher than 30-day mortality, whereas 90-day mortality was more than double the 30-day mortality rate. Cause of death data was analyzed, with the most frequent cause of death coded as malignant neoplasm of the esophagus for patients who died within 30 days of the operation (65.8%) and those who died between 30 and 90 days (58.7%; data not shown). Because these high numbers are likely the result of misunderstanding of International Classification of Diseases, 10th Revision, codes and miscoding, further analysis was not performed. Of those patients alive at 30 days, 7.7% (95% CI, 5.6% to 9.9%) died between 30 and 90 days. In our data set, too few deaths occurred within 30 days to generate robust predictive models. Hierarchical generalized linear regression models were, however, used to predict death within 90 days. The a priori variables chosen for this model were age, neoadjuvant therapy,

Number of Resections

30

20

10

0 0

50

100

150

Provider (Ranked by Volume)

Fig 2. Number of esophagectomies performed for each provider in the series. Each tick on the x-axis represents an individual provider, and each provider is ranked by total volume.

congestive heart failure, history of myocardial infarction, peripheral vascular disease, diabetes, and COPD. At 90 days, age and COPD were the only variables predictive of death in our data set (C statistic, 0.82; Table 3). In the 90-day model, provider covariance testing reached statistical significance (p ¼ 0.018), supporting the notion that variations in provider outcomes do exist. Removal of provider random effect and adjustment for patient-specific covariables demonstrated that only 3 of the 188 hospitals evaluated had an observed-to-expected (O/E) mortality rate different than 1 (p < 0.05), with all 3 hospitals exhibiting O/E ratios exceeding 1. However, when testing O/E ratios for 188 hospitals at a p value of less than 0.05, significant values are expected due to multiple testing. After Bonferroni adjustment for multiple comparisons, no hospitals had observed mortality rates different from expected.

Comment Thirty-day mortality after surgical procedures is collected and reported by a variety of administrative and national databases and is a frequently used surgical quality measure. However, because surgical and postoperative care has evolved, 30-day mortality currently provides only a snapshot of true surgical quality and significantly underestimates the true risk of certain operations. Our study supports this notion, because we have demonstrated that the mortality rate after esophagectomy in a population of Medicare patients doubles at 90 days (13.3%) compared with 30 days (6.0%). Unfortunately, comparison of this 90day mortality rate is difficult because previous series have generally reported only 30-day mortality. The 30-day mortality rate of 6.0% in the SEERMedicare data set is similar to many published series. However, specialized clinical databases have shown a significant reduction in death. For instance, Wright and colleagues [2] recently queried the STS-GTD and reported a 30-day mortality of 2.7% after esophagectomy. However, the STS database includes non-Medicare

4

WALTERS ET AL MORTALITY AFTER ESOPHAGECTOMY

Ann Thorac Surg 2014;-:-–-

Table 1. Patient Characteristics Variables Demographics Age, y 65–69 70–74 75–79 80–84 85 Sex Male Female Year of diagnosis 2006 2007 2008 2009 Race White Black Asian Hispanic Other Pathology Adenocarcinoma and variants SCC and variants Carcinoma, not otherwise specified Cancer stage 0 (tumor in situ) I II III IV Unknown Tumor stage Tis T1 T2 T3 T4 TX Nodal stage N0 N1 NX Induction therapy Chemotherapy Radiotherapy Chemotherapy or radiotherapy Comorbidities Myocardial infarction Heart failure Peripheral vascular disease Cerebral vascular disease COPD Diabetes Chronic renal failure COPD ¼ chronic obstructive pulmonary disease; cell carcinoma.

No. (%)

238 194 129 60 13

(37.5) (30.6) (20.4) (9.5) (2.1)

533 (84.1) 101 (15.9) 171 164 159 140

(27.0) (25.9) (25.1) (22.1)

584 24 8 5 13

(92.1) (3.8) (1.3) (0.8) (2.1)

479 (75.6) 145 (22.9) 10 (1.6) 10 165 218 182 32 21

(1.7) (27.2) (35.9) (30.0) (5.3) (3.6)

10 209 97 263 34 21

(1.6) (33.0) (15.3) (41.5) (5.4) (3.3)

332 (52.4) 292 (46.1) 10 (1.6) 180 (28.4) 265 (41.8) 289 (45.6) 41 47 26 16 155 130 28

(6.5) (7.4) (4.1) (2.5) (24.4) (20.5) (4.4)

SCC ¼ squamous

patients and principally comprises dedicated general thoracic surgeons at higher-volume centers, which may contribute to a lower mortality rate. In contrast, the SEER-Medicare database only captures patients aged older than 65 years. SEER-Medicare data may not be completely generalizable to all patients with esophageal cancer but may be more representative of current national trends for elderly patients because it includes all hospitals performing esophagectomy on Medicare patients. Interestingly, the number of esophagectomies performed per institution on Medicare patients in this study is low (ie, half of the institutions performed only 1). In addition, the mortality rate is alarmingly high, particularly when compared with the STS data. Although the patient populations between SEER-Medicare and STS are different, these observations pose the question of whether we are achieving, in our current national climate, optimal surgical outcomes after esophagectomy. The fact that 90-day mortality doubles compared with 30-day mortality has multiple clinical implications. First, it suggests that sequelae of an operation continue well beyond the 30-day postoperative period, which most surgeons would agree makes sense anecdotally. This particularly holds true for patients who have had a postoperative complication [2]. Accurate cause of death data were not available within this data set; however, that patients with esophageal cancer who undergo esophagectomy die of their cancer within 90 days is highly unlikely. Thus, it can be reasonably concluded that death within 90 days is a result of the surgical procedure rather than cancer progression. The second important clinical point is that we are underestimating the true operative risk in the Medicare population when counseling patients preoperatively. Better understanding of the true risk may lead to improvement of the informed consent process. Finally, understanding and recognizing that operative risk continues beyond 30 days may conceivably lead to improved long-term surveillance of patients after esophagectomy. With that recognition, we may then find ways for increased efficient use of adjunct health care resources (eg, physical therapy, nutritional support, and home health nurses) and, ultimately, improved postoperative survival. The multivariable regression model found only age and COPD were predictive of death at 90 days. Several authors have shown age is a risk factor for death after esophagectomy [16, 17], but others have refuted this [18–

Table 2. Postoperative Mortality Mortality

No. (%)

In-hospital 30-day Perioperative 90-day

50 38 54 84

(7.9) (6.0) (8.5) (13.3)

Ann Thorac Surg 2014;-:-–-

WALTERS ET AL MORTALITY AFTER ESOPHAGECTOMY

Table 3. Hierarchical Linear Regression Model: 90-Day Mortality Variable Age Induction therapy Congestive heart failure History of myocardial infarction Peripheral vascular disease Diabetes mellitus COPD a

OR

95% CI

p Value

F Test

1.76 0.81 1.05 1.68

1.13–2.75 0.49–1.34 0.45–2.48 0.70–4.03

0.01 0.68 0.90 0.25

6.31 0.68 0.01 1.35

1.04 0.83 1.79

0.33–3.23 0.43–1.57 1.06–3.00

0.95 0.56 0.03

0.00 0.34 4.85

a

Odds ratio for age in 10-year increments.

CI ¼ confidence interval; COPD ¼ chronic obstructive pulmonary disease; OR ¼ odds ratio.

20]. However, many of the studies refuting age as a risk factor for death analyzed age as a categoric variable, arbitrarily choosing 70 years as a cutoff. Age is a continuous variable, and when we analyzed it as such, we demonstrated a significant association between advanced age and death after esophagectomy when controlling for other variables, including comorbid disease. Pulmonary complications are the most frequent complications after esophagectomy, occurring in 16% to 27% of patients [17, 21], and often lead to poor outcomes. In fact, Law and colleagues [17] demonstrated that 55% of in-hospital deaths after esophagectomy are the result of pulmonary complications. Similar to our series, other authors have shown preoperative pulmonary disease is a risk factor for postoperative pulmonary complications [21]. Our study calls into question whether postoperative mortality is a useful quality metric after esophagectomy. Only 3 of the 188 hospitals in this study sample had O/E mortality ratios that were statistically different than 1 at a p value of 0.05. However, after adjustment for multiple comparisons, no hospitals had rates that were convincingly different than the expected mortality rate. This is not surprising, given the relatively low frequency of mortality events and the relative low numbers of operations performed at each institution. Dimick and colleagues [22] examined surgical mortality as a quality indicator for a variety of procedures, including esophagectomy, using the Nationwide Inpatient Sample. They concluded that for esophagectomy, detection of a doubling of expected mortality rate would require a minimum caseload of 77 patients. They determined that only 1% and 4% of institutions nationally would meet this minimum requirement at 3 years and 5 years, respectively. For detection of more subtle differences, such as a 1.5-times increase over the benchmark, no institutions would meet criteria, even when including 5 years of data. Given this evidence in combination with our data, even 90-day mortality is a poor quality metric because it does not distinguish between hospitals that perform well and those that perform poorly. This study has several limitations. The patient population within the SEER-Medicare database only includes

5

patients aged older than 65 years, and thus, these results may not be applicable to the entire population of patients undergoing esophagectomy. However, the median age of patients in our study was 72 years, which is older but not greatly different from the reported median age of all patients with esophageal cancer of 67 years [23]. Small sample size limited additional conclusions regarding risk factors for death at 30 days and 90 days. Incorporating additional years of data may improve these analyses; however, as more years are added to the data set, the analyses become increasingly limited by evolving operative techniques and postoperative care strategies. The small sample size is worth comment, because 89.7% of patients within the complete data set did not undergo esophagectomy. That at least some of these patients had surgically resectable disease is highly likely, and this may underscore a bias within the medical community against referring elderly patients for esophagectomy. Finally, this study was not able to assess the effect of different approaches to esophagectomy (ie, Ivor-Lewis, transhiatal approaches), determine the effect of minimally invasive esophagectomy on this patient population, or make comparisons between esophagectomy and endoscopic mucosal resection for early-stage cancers. Comparison between types of esophagectomy is made difficult because it is limited by International Classification of Diseases, 9th Revision, procedural codes and because nearly two-thirds of patients had an unspecified approach in this study. Clearly, more work needs to be done to fully understand the effect of surgical approach on 90-day mortality. As surgeons, we are rightfully held accountable for postoperative complications and death. In the current and future climate of quality-improvement initiatives, value-based purchasing, and fee for service, we can expect to be judged for the quality of care we provide our patients. If we are to be held accountable for our outcomes, we need to choose relevant outcomes, accurately define them, and properly risk adjust to allow for comparison among surgeons and institutions. As we have demonstrated here, 90-day mortality more accurately reflects the true operative risk of esophagectomy in an elderly Medicare patient population. Unfortunately, even when considering 90-day mortality as a quality measure, it does not differentiate good hospitals from hospitals those that perform poorly. However, accurately reporting a patient’s true postoperative mortality risk after esophagectomy can improve the informed consent process and may affect patient selection for the operation. This research was funded by the Agency for Healthcare Research and Quality grant K08-HS18049 to B.D.K.

References 1. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer J Clin 2013;63:11–30.

6

WALTERS ET AL MORTALITY AFTER ESOPHAGECTOMY

2. Wright CD, Kucharczuk JC, O’Brien SM, Grab JD, Allen MS; Society of Thoracic Surgeons General Thoracic Surgery Database. Predictors of major morbidity and mortality after esophagectomy for esophageal cancer: a Society of Thoracic Surgeons General Thoracic Surgery Database risk adjustment model. J Thorac Cardiovasc Surg 2009;137:587–96. 3. Mamidanna R, Bottle A, Aylin P, Faiz O, Hanna GB. Shortterm outcomes following open versus minimally invasive esophagectomy for cancer in England: a population-based national study. Ann Surg 2012;255:197–203. 4. Chang AC, Ji H, Birkmeyer NJ, Orringer MB, Birkmeyer JD. Outcomes after transhiatal and transthoracic esophagectomy for cancer. Ann Thorac Surg 2008;85:424–9. 5. Zimmerman JE, Kramer AA, Knaus WA. Changes in hospital mortality for United States intensive care unit admissions from 1988 to 2012. Crit Care 2013;17:R81. 6. Mayo SC, Shore AD, Nathan H, et al. Refining the definition of perioperative mortality following hepatectomy using death within 90 days as the standard criterion. HPB (Oxford) 2011;13:473–82. 7. Visser BC, Keegan H, Martin M, Wren SM. Death after colectomy: it’s later than we think. Arch Surg 2009;144:1021–7. 8. Bryant AS, Rudemiller K, Cerfolio RJ. The 30- versus 90-day operative mortality after pulmonary resection. Ann Thorac Surg 2010;89:1717–22. 9. Hu Y, McMurry TL, Wells KM, Isbell JM, Stukenborg GJ, Kozower BD. Postoperative mortality is an inadequate quality indicator for lung cancer resection. Ann Thorac Surg 2014;97:973–9. 10. LaPar DJ, Stukenborg GJ, Lau CL, Jones DR, Kozower BD. Differences in reported esophageal cancer resection outcomes between national clinical and administrative databases. J Thorac Cardiovasc Surg 2012;144:1152–7. 11. Warren JL, Klabunde CN, Schrag D, Bach PB, Riley GF. Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population. Med Care 2002;40(8 Suppl):3–18.

Ann Thorac Surg 2014;-:-–-

12. Klabunde CN, Potosky AL, Legler JM, Warren JL. Development of a comorbidity index using physician claims data. J Clin Epidemiol 2000;53:1258–67. 13. National Cancer Institute. 2010 SEER-Medicare: calculation of comorbidity weights. Available at: http://healthservices. cancer.gov/seermedicare/program/comorbidity.html, 2013. Accessed Jan 24, 2014. 14. R Development Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2011. 15. Wickham H. Ggplot2: elegant graphics for data analysis. 2nd edition. New York, NY: Springer; 2009. 16. Ferguson MK, Martin TR, Reeder LB, Olak J. Mortality after esophagectomy: risk factor analysis. World J Surg 1997;21: 599–603. 17. Law S, Wong KH, Kwok KF, Chu KM, Wong J. Predictive factors for postoperative pulmonary complications and mortality after esophagectomy for cancer. Ann Surg 2004;240: 791–800. 18. Ellis FH Jr, Williamson WA, Heatley GJ. Cancer of the esophagus and cardia: does age influence treatment selection and surgical outcomes? J Am Coll Surg 1998;187:345–51. 19. Poon RT, Law SY, Chu KM, Branicki FJ, Wong J. Esophagectomy for carcinoma of the esophagus in the elderly: results of current surgical management. Ann Surg 1998;227: 357–64. 20. Ruol A, Portale G, Zaninotto G, et al. Results of esophagectomy for esophageal cancer in elderly patients: age has little influence on outcome and survival. J Thorac Cardiovasc Surg 2007;133:1186–92. 21. Avendano CE, Flume PA, Silvestri GA, King LB, Reed CE. Pulmonary complications after esophagectomy. Ann Thorac Surg 2002;73:922–6. 22. Dimick JB, Welch HG, Birkmeyer JD. Surgical mortality as an indicator of hospital quality: the problem with small sample size. JAMA 2004;292:847–51. 23. Enzinger PC, Mayer RJ. Esophageal cancer. N Engl J Med 2003;349:2241–52.

DISCUSSION DR DAVID TOM COOKE (Sacramento, CA): Thank you. Very fascinating presentation. I am going to ask a quick question, and then we’ll move forward with other questions in the audience. So honestly, there is a fair amount of limitations with the Surveillance, Epidemiology, and End Results (SEER) database. It is a very good database, but it lacks some key elements. Two questions I have: One, I notice your database allows you to look at the ZIP code for the treating institution, but does it also provide you with geographic data for the patient that was treated? For instance, those patients who had 90-day mortality vs those patients who died in the hospital, was it a matter of the hospital keeping these patients around because they know they would have to travel far and are from a rural area, or were there real geographic differences? And the second thing is, I notice your database does not provide you with a discharge location, for instance, discharge to an institutional care facility. Have you had thoughts about taking this research further to another database, say, the Nationwide Inpatient Sample, to see if those patients with 90-day mortality were more likely to be discharged to an institutional care facility versus home? DR WALTERS: Those are both excellent questions. With regard to your first question, unfortunately, SEER does not allow for analysis regarding how far patients are traveling for their procedures and whether patients remain in the hospital due to geographical challenges, but that is a very important point. And your second question, remind me?

DR COOKE: Discharge to an institutional care facility versus home. DR WALTERS: Yes, another great question. Some patients may be discharged to skilled nursing or rehab facilities due to concern regarding their postoperative survival or well-being, but SEER does not allow us to tease that information out. It would be very helpful to know where these patients are going and correlate that to outcomes. It is also possible that those that are going to rehabs and nursing facilities are getting closer follow-up and more attention. DR COOKE: Or are they sicker, and more likely to die? DR WALTERS: That is true. DR MARK B. ORRINGER (Ann Arbor, MI): Dr Walters, I congratulate you and your colleagues. I think that the 90-day mortality after esophagectomy is an extremely important concept. And in The Society of Thoracic Surgeons (STS), we need to take the lead in identifying and reporting the true mortality after esophagectomy. A patient I saw 3 days ago had a minimally invasive esophagectomy performed by a less experienced surgeon of the type to whom you alluded. The intrathoracic anastomosis fell apart on day 5 or 6, the gastric conduit was resected, the empyema drained, a cervical esophagostomy established, and he spent

Ann Thorac Surg 2014;-:-–-

4 1/2 months in the intensive care unit at the other hospital with a tracheostomy, respiratory insufficiency, and sepsis. He now presents to me 6 months later and will have to undergo a substernal colon interposition to reestablish alimentary continuity. If he dies perioperatively after his colon interposition, this is really an operative mortality of the esophagectomy that was done months ago. We have to have a better way of documenting such adverse long-term outcomes of these operations. Ninety-day mortality data after esophagectomy are certainly more meaningful than “hospital mortality” alone. Operative mortality after esophagectomy has been related to hospital and surgeon volume. Obtaining good results after a traditional open esophagectomy are hard enough. An even smaller number of surgeons are facile in minimally invasive esophagectomy, and an even smaller number in robotic esophagectomy. These latter minimally invasive techniques are being touted by some as the new “gold standard” for esophagectomy. Life is not so simple. The complications we are seeing from these procedures are huge and can lead to very late mortality. I think you have highlighted an important new data point for documentation in the STS General Thoracic Surgery Database. I commend you for doing this. DR WALTERS: Thank you. DR WAYNE HOFSTETTER (Houston, TX): Very nice talk. I have to tell you that I am very troubled by the data. And I am going to accuse you of something that I am guilty of as well, which is that in our introductions, we spend a lot of time talking about the poor outcomes of esophagectomy and the high morbidity and mortality. Basically, this overall impression will be used against us by people who are really pushing definitive chemoradiation as the treatment of choice for esophageal cancer. And I do not mean just squamous cell. I am talking about adenocarcinoma, as well. So number one, I would urge all of us, myself included, to be honest in our writing but also to highlight the data that exist right

WALTERS ET AL MORTALITY AFTER ESOPHAGECTOMY

7

now in centers of excellence and not to recapitulate the poor outcomes of the past. This method of emphasizing the worst outcomes to give the reader a sense of urgency is backfiring; patients and referring physicians have access to this research on the Internet. Having said that, I will comment that I think the SEER database 30-day and 90-day mortalities are excessive. And I do not mean to preach to the choir here, but those numbers should be, in our hands, around 2% at 30 days and 4% to 5% at 90 days. When we get to rates like 13% quoted in this data, this is the reason that we are losing out to definitive chemoradiation. Such was the case in the European trials of definitive chemoradiation versus chemoradiation (CXRT) plus surgery, where they lost 12% to 14% of their postoperative patients. That excess mortality takes away any potential benefit gained from surgery. What am I saying? I am saying that centers of excellence and regionalization is where we need to go. We need to get the message out that those kinds of mortalities are not acceptable, because that is not the benchmark that we need to be measuring ourselves against. DR WALTERS: I think that is an excellent point, and I think that highlights an opportunity for the STS. If you look at Dr Wright and Dr Allen’s paper published in the Journal of Thoracic and Cardiovascular Surgery (JTCVS) in 2009, they report a 2.7% inhospital mortality rate after esophagectomy, using the STS database. The SEER database certainly demonstrates a much higher operative mortality. And this is a big opportunity, I think, for thoracic surgeons to say that we do this operation with a very low mortality rate in a safe manner; and across the country, that is not necessarily the case. DR HOFSTETTER: I would really encourage you to put that in the discussion of your paper. DR WALTERS: Thanks.

Understanding mortality as a quality indicator after esophagectomy.

Postoperative death is an important outcome after esophagectomy, and the Centers for Medicare & Medicaid Services currently uses 30-day mortality as a...
234KB Sizes 0 Downloads 3 Views