Abstract
Background and Objectives
Epidural analgesia may increase survival after cancer surgery by reducing recurrence. This population-based study compared survival and treated recurrence after gastric cancer resection between patients receiving epidurals and those who did not.
Methods
We used the linked federal Surveillance, Epidemiology, and End Results (SEER) Program/Medicare database to identify patients aged ≥66 with nonmetastatic gastric carcinoma diagnosed 1996–2005 who underwent resection. Exclusions included diagnosis at autopsy, no Medicare Part B, familial cancer syndrome, emergency surgery, and laparoscopic procedures. Epidurals were identified by CPT codes.
Treated recurrence was defined as chemotherapy ≥ 16 months and/or radiation ≥ 12 months after surgery. Recurrence was compared by conditional logistic regression. Survival was compared via marginal Cox proportional hazards regression.
Results
We identified 2745 patients, 766 of whom had epidural codes. Patients receiving epidurals were more likely to have regional disease, be white, and live in areas with relatively high socioeconomic status. Overall treated recurrence was 25.6% (27.5% epidural and 24.9% non-epidural). In the adjusted logistic regression, there was no difference in recurrence (odds ratio [OR] 1.40, 95% CI 0.96 to 2.05).
Median survival did not differ: 28.1 months (95% CI 24.8 to 32.3) in the epidural versus 27.4 months (95% CI 24.8 to 30.0) in the non-epidural groups. The marginal Cox models showed no association between epidural use and mortality (adjusted HR 0.93, 95% CI 0.84 to 1.03).
Conclusions
There was no difference between groups regarding treated recurrence or survival. Whether this is true or simply a result of insufficient power is unclear. Prospective studies are needed to provide stronger evidence.
INTRODUCTION
Although much less common than other malignancies, gastric cancer carries a grave prognosis. In 2013, the American Cancer Society estimated a total of 21,600 people would be diagnosed with gastric cancer and an estimated 10,990 patients would die from it.1 A key determinant of outcome in patients operated for gastric cancer is local or regional recurrence,2 with reported rates after resection with curative intent of 21% to 49%.3–4 Even with optimal surgical technique, tumor surgery may release tumor cells into the lymphatics and vasculature, and a large fraction of patients already harbor remote foci of tumor cells at the time of surgery.5–6
Whether these factors result in clinically apparent disease depends largely on the balance between immune activity (tumor surveillance) and the tumor’s ability to invade, proliferate, and promote angiogenesis.7 Realistically, the immune system and other host defenses frequently fail to eliminate minimal residual disease. As a result, local recurrence and metastatic disease remain common after gastric cancer surgery.
Several perioperative factors have been found to affect the balance between metastasis and tumor surveillance in the perioperative period.5–6,8–11 General anesthesia,12–13 opioid analgesics,14–15 and allogeneic blood transfusion, which is often necessary in large surgical procedures, are known to be immunosuppressive. Regional or neuraxial anesthetic techniques, even when combined with general anesthesia, may suppress immune function less than opioid analgesia15–16 by reducing the surgical stress response and significantly reducing opioid exposure. For example, regional anesthetic techniques have been associated with lower recurrence rates of breast and prostate cancers.17–18 However, results for colon cancer vary,19–21 while the literature is silent on gastric cancer.
The effect of neuraxial anesthetic techniques on survival after surgery has also been heavily debated. Although original reports suggested a survival benefit of neuraxial anesthesia/analgesia for different types of surgery,22 published results for survival after colorectal cancer surgery have differed, with some studies demonstrating a survival benefit19,23–24 and others finding none.21 While cancer recurrence will impact survival to a large extent, neuraxial anesthesia/analgesia may also reduce perioperative cardiac and respiratory events. 22
Although retrospective by nature, administrative databases provide data not easily found outside of large randomized clinical trials.25 In particular, Medicare claims provide a large volume of patient encounters for research. Using the linked Surveillance, Epidemiology, and End Results (SEER)-Medicare database, we compared overall survival and cancer recurrence rates between patients who did or did not receive epidural anesthesia and/or analgesia for open resection of nonmetastatic gastric cancer. Our hypotheses were that epidural anesthesia and/or analgesia are associated with (1) reduced cancer recurrence after gastric cancer resection and (2) improved all-cause mortality after surgery.
METHODS
This study was approved by the National Cancer Institute and the Institutional Review Boards of Cleveland Clinic and University Hospitals Case Medical Center, Cleveland, Ohio.
Data Sources
The SEER tumor registry of the National Cancer Institute tracks cancer incidence in a set of cancer registries across the United States which currently encompass over a quarter of the U.S. population. Data collected include patient demographics (age, race, gender, marital status), primary site, prior cancer diagnoses, histology, stage, site specific surgery, and initial course of therapy. Patients in the SEER database who are also enrolled in Medicare can be identified by linking SEER and Medicare data, resulting in a joint database of patients aged ≥ 65 years who were diagnosed with cancer while living in one of the SEER areas. The Patient Entitlement and Diagnosis Summary File (PEDSF) includes data from SEER and Medicare enrollment data. Medicare Part A claims for inpatient hospital care are contained in the Medicare Provider and Analysis Review (MedPAR) file. Medicare Part B claims for physician services and outpatient services are available in the Outpatient Standard Analytical File and the National Claims History files. International Classification of Diseases, 9th Edition-Clinical Modification (ICD-9-CM) and Current Procedural Terminology, 4th Edition (CPT-4) procedure codes can be used to identify procedures from Medicare hospital inpatient claims (MedPAR) and physician-supplier claims. The SEER-Medicare database and its use in cancer research have been described previously.26
Study Population
Patients aged ≥ 66 years diagnosed with incident non-metastatic gastric carcinoma between 1996 and 2005 (identified from PEDSF) who underwent gastrectomy (identified from MedPAR) were included. Cases with anatomical site codes for fundus, body, gastric antrum, pylorus, lesser and greater curvature, overlapping lesion, and stomach were included. The most recent available Medicare data were from 2009, providing at least 4 years of follow-up for cancer recurrence. Patients aged 66 years and older were included to allow for available claims to assess comorbid conditions 1 year prior to gastric cancer diagnosis. Using SEER Summary Stage, patients who presented with localized or regional stage disease and underwent gastrectomy within 6 months of diagnosis as indicated in Medicare claims data were included.
The following cases (identified in PEDSF or from Medicare data) were excluded: carcinoma in situ, metastatic disease, or unstaged cancers; prior diagnosis of cancer; cancer diagnosed on autopsy or death certificate only; eligibility for Medicare on the basis of end stage renal disease or disability; and patients with Medicare Health Maintenance Organization enrollment or without both Medicare Parts A and B coverage, due to lack of claims data needed for the analysis. Patients undergoing laparoscopic gastrectomy were excluded due to low likelihood of epidural catheter placement. Patients with diagnoses suggesting urgent or emergent procedures according to the method of Diggs et al27 were excluded. Finally, patients who developed an unrelated second malignancy (indicated in PEDSF) after gastric cancer diagnosis were excluded.
Using these criteria, 2 cohorts were identified. To ensure complete claims history, the cohort for assessment of survival consisted of patients enrolled in Medicare Part A and B within 1 year prior to cancer diagnosis until death or 8 months after diagnosis, whichever came first, in order to capture perioperative data and complications. The cohort for assessment of cancer recurrence consisted of patients enrolled in Medicare Part A and B within 1 year prior to cancer diagnosis until death or 4 years after diagnosis who survived at least 12 months after surgery. Figure 1 details the application of the exclusion criteria.
Figure 1.
Application of exclusion criteria
Measures
Exposure
CPT codes for placement of thoracic and lumbar epidural catheters (62318 and 62319, respectively) and daily hospital management of an epidural infusion (01996) were identified from the Standard Analytical File and National Claims History file. Epidural codes were identified ranging from one week prior to and after surgery. These codes have been previously been used successfully in Medicare database research.19,28–30 Cases with codes for epidural placement ordaily management were considered to be in the epidural group.
Outcome
All-cause mortality after cancer resection was determined by the Medicare date of death as documented in PEDSF. The secondary outcome was treated gastric cancer recurrence, which was defined within a 4-year window after surgery as an ICD-9 metastasis code (197.5) or receipt of chemotherapy ≥ 16 months after resection and/or radiation therapy ≥ 12 months after resection.31–32 This approach has been validated using SEER-Medicare data in a previous study.33
Covariates
From PEDSF, patient-level demographic and clinical data were collected including age, gender, race, marital status, SEER site, year of diagnosis, anatomical site, and cancer stage at diagnosis. Census tract level variables (urban/rural residence, median income, and percent of residents with at least a high school diploma) were used as a proxy for socioeconomic status. Blood transfusions were noted from MedPAR claims.
Comorbid conditions were identified using ICD-9-CM codes from MedPAR, Outpatient, and physician-supplier claims to calculate a comorbidity score modified from the Deyoadaptation of the Charlson comorbidity index34 as recommended by the National Cancer Institute.* Klabunde’s 30-day rule out algorithm was applied to incorporate comorbid conditions from outpatient claims.35
Gastrectomy
We examined MedPAR and physician-supplier claims for surgical procedure codes. We used the following ICD-9-CM procedure codes for gastrectomy: proximal gastrectomy (43.5), partial gastrectomy with anastomosis to duodenum (43.6), partial gastrectomy with anastomosis to jejunum (43.7), other partial gastrectomy (43.8x), and total gastrectomy (43.9x). Additionally, the following CPT codes were evaluated: total gastrectomy (43620-22) and partial distal gastrectomy (43631-43634). The use of an ICD-9-CM diagnosis code of V64.4 (conversion of a laparoscopic procedure to open) in conjunction with procedure codes for gastrectomy in the absence of procedure codes for laparoscopy (54.21, 45.81) was considered to be a laparoscopic procedure that was converted to open. These patients were excluded from the analytic cohort. If a patient had a gastrectomy and the codes for laparoscopy were not present, then the patient was assumed to have had an open gastrectomy.
Complications
Medicare claims data from surgery date until 30 days after surgery were evaluated for perioperative complications. Complications included ICD-9 or CPT-4 codes for retrieval of retained foreign body, accidental puncture or laceration, incision of abdominal wall, management of postoperative shock/hemorrhage, management of abdominal infection, repair of an organ injury/laceration, reoperative laparotomy, management of wound complication, management of fistula, and management of bowel obstruction.36
Statistical Analysis
Patient characteristics were compared between the epidural anesthesia and/or analgesia group and the non-epidural group. For categorical variables, Pearson’s chi square test was used. For continuous variables, medians with interquartile ranges are presented. Wilcoxon rank sum tests were used to compare the groups.
Survival time was measured from the date of gastrectomy to death or last follow-up through December 31, 2009, indicated as censor. Kaplan-Meier survival curves were generated. Survival times between treatment groups (epidural versus no epidural) were compared using the log-rank test. For multivariate survival analysis, a marginal Cox model was constructed given the nature of the hospital clustered data. This method yields a stronger estimate of covariate effects. A robust sandwich estimate was implemented in the model.37 As a confirmatory analysis, using logistic regression, a propensity score was generated to predict the probability of the use of epidural anesthesia using the same covariates. The propensity score was then incorporated into the survival models as a continuous covariate.
In the analysis of the recurrence cohort, a conditional logistic regression was constructed to predict the likelihood of treated cancer recurrence, controlling for hospital effect.
Finally, in an effort to determine what treatment effect would be detectable given these results, an exploratory post hoc power analysis was conducted using the observed cohort size and proportion of epidural use.
Analysis was conducted with SAS software version 9.3 (SAS Institute, Cary, NC). All comparisons used 2-sided tests with P values < 0.05 considered significant.
RESULTS
Patient Selection
Using the inclusion and exclusion criteria described, we identified a cohort of 2745 patients, of whom 27.9% (n=766) had epidurals. The cohort for analysis of cancer recurrence comprised 1882 patients, of whom 28.8% (n=542) had epidurals at the time of resection.
Patient Characteristics
There were a number of differences between the 2 groups (Table 1). Patients who received epidurals were more likely to be white and live in areas with higher median income and education. Epidural use was also more common in the Midwest SEER region and increased slightly over the study interval. Epidural use differed by tumor extent as well, with patients receiving epidurals more likely to have regional stage disease. Epidural use did not differ significantly by tumor location in the stomach. Patients with epidurals were also slightly more likely to have surgical complications.
Table 1.
Patient Characteristics by Type of Pain Management
| Characteristics | Total sample (n=2745) | Traditional Pain Management (n=1979) | Epidural Anesthesia and/or Analgesia (n=766) | P Value |
|---|---|---|---|---|
|
| ||||
| Age at diagnosis (95% CI) | 76.8 (72.3, 82.2) | 76.9 (72.5, 82.3) | 76.5 (72.0, 81.8) | 0.0709 |
|
| ||||
| Categorical age | ||||
| 66–69 | 397 (14.46) | 273 (13.79) | 124 (16.19) | 0.3338 |
| 70–74 | 683 (24.88) | 488 (24.66) | 195 (25.46) | |
| 75–79 | 736 (26.81) | 531 (26.83) | 205 (26.76) | |
| 80–84 | 555 (20.22) | 405 (20.46) | 150 (19.58) | |
| 85+ | 374 (13.62) | 282 (14.25) | 92 (12.01) | |
|
| ||||
| Gender | ||||
| Female | 1323 (48.20) | 972 (49.12) | 351 (45.82) | 0.1214 |
| Male | 1422 (51.80) | 1007 (50.88) | 415 (54.18) | |
|
| ||||
| Race | ||||
| White | 1534 (55.88) | 1020 (51.54) | 514 (67.10) | <0.0001 |
| Black | 288 (10.49) | 219 (11.07) | 69 (9.01) | |
| Other | 923 (33.62) | 740 (37.39) | 183 (23.89) | |
|
| ||||
| Marital status | ||||
| Married | 1509 (54.97) | 1071 (54.12) | 438 (57.18) | 0.3406 |
| Not married | 1157 (42.15) | 849 (42.90) | 308 (40.21) | |
| Unknown | 79 (2.88) | 59 (2.98) | 20 (2.61) | |
|
| ||||
| Education* (95% CI) | 81.0 (69.8, 89.1) | 79.6 (67.5, 88.8) | 83.6 (74.4, 90.0) | <0.0001 |
|
| ||||
| Income† | 42454 (31067, 57433) | 41895 (30361, 57280) | 44647 (32850, 59290) | 0.0017 |
|
| ||||
| Residence | ||||
| Metropolitan | 2439 (88.85) | 1766 (89.24) | 673 (87.86) | 0.4918 |
| Urban | >270 (>9.8) | 190 (9.60) | 85 (11.10) | |
| Rural | ** | ** | ** | |
|
| ||||
| SEER region | ||||
| Midwest | 371 (13.52) | 180 (9.10) | 191 (24.93) | <0.0001 |
| Northeast | 569 (20.73) | 423 (21.37) | 146 (19.06) | |
| South | 329 (11.99) | 250 (12.63) | 79 (10.31) | |
| West | 1476 (53.77) | 1126 (56.90) | 350 (45.69) | |
|
| ||||
| Charlson Index | ||||
| 0 | 1970 (71.77) | 1405 (71.00) | 565 (73.76) | 0.0626 |
| 1 | 505 (18.40) | 361 (18.24) | 144 (18.80) | |
| 2 | 164 (5.97) | 127 (6.42) | 37 (4.83) | |
| 3+ | 106 (3.86) | 86 (4.35) | 20 (2.61) | |
|
| ||||
| Tumor differentiation | ||||
| Well | 142 ((5.17) | 115 (5.81) | 27 (3.52) | 0.1109 |
| Moderately | 849 (30.93) | 610 (30.82) | 239 (31.20) | |
| Poorly | 1576 (57.41) | 1125 (56.85) | 451 (58.88) | |
| Others/Unknown | 178 (6.48) | 129 (6.52) | 49 (6.40) | |
|
| ||||
| Blood transfusion | ||||
| No | 2370 (86.34) | 1708 (86.31) | 662 (86.42) | 0.9363 |
| Yes | 375 (13.66) | 271 (13.69) | 104 (13.58) | |
|
| ||||
| Stage | ||||
| Local | 999 (36.39) | 750 (37.90) | 249 (32.51) | 0.0085 |
| Regional | 1746 (63.61) | 1229 (62.10) | 517 (67.49) | |
|
| ||||
| Year of diagnosis | ||||
| 1996 | 177 (6.45) | 121 (6.11) | 56 (7.31) | 0.0082 |
| 1997 | 189 (6.89) | 120 (6.06) | 69 (9.01) | |
| 1998 | 174 (6.34) | 121 (6.11) | 53 (6.92) | |
| 1999 | 201 (7.32) | 130 (6.57) | 71 (9.27) | |
| 2000 | 319 (11.62) | 236 (11.93) | 83 (10.84) | |
| 2001 | 349 (12.71) | 257 (12.99) | 92 (12.01) | |
| 2002 | 341 (12.42) | 253 (12.78) | 88 (11.49) | |
| 2003 | 306 (11.15) | 237 (11.98) | 69 (9.01) | |
| 2004 | 368 (13.41) | 262 (13.24) | 106 (13.84) | |
| 2005 | 321 (11.69) | 242 (12.23) | 79 (10.31) | |
|
| ||||
| Cancer site | ||||
| Fundus | 108 (3.93) | 73 (3.69) | 35 (4.57) | 0.1086 |
| Body | 303 (11.04) | 219 (11.07) | 84 (10.97)) | |
| Antrum | 966 (35.19) | 727 (36.74) | 239 (31.20) | |
| Pylorus | 175 (6.38) | 126 (6.37) | 49 (6.40) | |
| Lesser curvature | 449 (16.36) | 315 (15.92) | 134 (17.49) | |
| Greater curvature | 191 (6.96) | 126 (6.37) | 65 (8.49) | |
| Overlapping lesion | 226 (8.23) | 155 (7.83) | 71 (9.27) | |
| Stomach, NOS | 327 (11.91) | 238 (12.03) | 89 (11.62) | |
|
| ||||
| Complications | ||||
| No | 2449 (89.22) | 1780 (89.94) | 669 (87.34) | 0.0482 |
| Yes | 296 (10.78) | 199 (10.06) | 97 (12.66) | |
Percent of residents in census tract with at least a high school education
Counts < 11 masked (and one adjacent cell with reduced precision) due to National Cancer Institute privacy requirements
Median income (U.S. dollars) of census tract
Association Between Epidural use and Overall Survival
Median survival in the epidural group was 28.1 months (95% CI 24.8 – 32.3); in the non-epidural group, median survival was 27.4 months (95% CI 24.8 – 30.0). Figure 2 shows Kaplan-Meier survival curves for the two groups; the curves were not significantly different (log-rank test P value= 0.54). Table 2 presents both the unadjusted and adjusted marginal Cox models. Model checking was based on cumulative sums of Martingale based residuals and the models had a good fit. Adjusting for multiple patient characteristics, there was no significant association between epidural use and improved overall survival (adjusted hazard ratio (HR) = 0.93, 95%CI 0.84 – 1.03;P=0.17). Covariates independently associated with increased mortality hazard include increasing age, male gender, surgical complications, increasing Charlson score, increasing tumor grade and stage, tumor location, and blood transfusion. For transfusion, the adjusted hazard ratio for mortality was 1.18, 95% CI 1.02 to 1.36;P value = 0.026.
Figure 2.
Kaplan-Meier survival curves for patients with and without epidural billing codes
Table 2.
Association Between Type of Pain Management and All-Cause Mortality
| Models* | Hazard Ratio | 95% CI | P Value |
|---|---|---|---|
|
| |||
| Model 1A | 0.97 | (0.87, 1.09) | 0.65 |
| Model 1B | 0.93 | (0.84, 1.03) | 0.17 |
| Model 2A | 0.95 | (0.85, 1.07) | 0.39 |
| Model 2B | 0.95 | (0.85, 1.07) | 0.36 |
| Model 2C | 0.94 | (0.84, 1.06) | 0.26 |
Marginal clustered survival models based on robust sandwich estimate
Model 1A – Epidural alone (unadjusted)
Model 1B – Epidural adjusted by age, gender, race, marital status, education, median income, year of diagnosis, Charlson score, tumor grade, stage, blood transfusion, cancer site and surgical complications.
Model 2A – Epidural adjusted by propensity score
Model 2B – Epidural adjusted by propensity score and blood transfusion
Model 2C – Epidural adjusted by propensity score and surgical complications
(Propensity score model includes age, gender, race, marital status, education, median income, year of diagnosis, Charlson score, tumor grade, stage, cancer site, SEER regions and residence type. Transfusion and surgical complications were not included because they would not logically predict the use of epidural analgesia.)
As a confirmatory analysis, we used the same covariates used in the adjusted Cox model (omitting blood transfusion and postoperative complications because they would not predict the use of epidural analgesia) to create a logistic regression model predicting a propensity score (likelihood of receiving an epidural). We then incorporated this propensity score as a covariate in the adjusted Cox proportional hazards analysis (Table 2, Models 2A, 2B, and 2C). Adjusting for the propensity score as a continuous variable, the results did not change significantly (HR=0.95, 95% CI 0.85 – 1.07; P=0.32).
Association Between Epidural Use and Treated Cancer Recurrence
Overall 4-year treated recurrence was 25.6% (27.5% in the epidural group and 24.9% in the non-epidural group. In the unadjusted conditional logistic regression (Table 3), there was no significant association between epidural use and cancer recurrence (OR = 1.38, 95%CI 0.98 – 1.95; P value = 0.06). There were, however, a number of significant predictors of treated recurrence: male gender, increasing Charlson score, higher tumor grade, increasing stage, residence in an area with higher education, and later year of diagnosis were associated with higher odds of recurrence. Adjusting for these and other demographic and clinical covariates did not appreciably change the result describing the association between epidural use and treated tumor recurrence (OR = 1.40, 95% CI 0.96 –2.05; P value = 0.08).
Table 3.
Association Between Type of Pain Management and Tumor Recurrence
| Models* | Odds Ratio | 95% CI | P Value |
|---|---|---|---|
|
| |||
| Model 1A | 1.38 | (0.98, 1.95) | 0.06 |
| Model 1B | 1.40 | (0.96, 2.05) | 0.08 |
| Model 1C | 1.37 | (0.97, 1.93) | 0.07 |
| Model 1D | 1.38 | (0.98, 1.94) | 0.07 |
Conditional logistic regression models adjusting for hospital effect.
Model 1A – Epidural alone (unadjusted)
Model 1B – Epidural adjusted by age, sex, race, marital status, education, median income, year of diagnosis, blood transfusion, Charlson score, tumor grade, stage, cancer site and surgical complications.
Model 1C – Epidural adjusted by surgical complications only (Surgical complication is not predictive of recurrence in the adjusted model)
Model 1D – Epidural adjusted by blood transfusion only (Blood transfusion is not predictive of recurrence in the adjusted model).
Post-Hoc Power Analysis
Assuming the worst case of 766 per group with 24.9% treated recurrence in the non-epidural group, this analysis yielded 86% power to detect a 7% absolute difference in treated recurrence and 93% power to detect an 8% difference. These relatively small differences suggest adequate statistical power (within the limitations of a post-hoc analysis).
DISCUSSION
Given the preclinical and observational data supporting immune function as a protective factor against cancer spread, it is reasonable to study whether interventions aimed at reducing exposure to immunosuppressive factors would improve patient outcomes after cancer resection. This has, however, been difficult to demonstrate with available data for gastric cancer. Unlike prior observational studies evaluating colorectal,19 breast,17 and prostate18 cancer, this analysis failed to find an association between epidural use and either treated recurrence or differences in mortality after gastrectomy.
There is literature investigating other malignancies that is consistent with our findings. Gottschalk et al. analyzed 669 patients undergoing colorectal cancer resection and found no overall association between epidural analgesia and recurrence risk, although their study was limited by length of follow-up.20 Myles et al21 evaluated recurrence of multiple abdominal cancers in groups randomized to either epidural or systemic analgesia. They found no significant difference in overall recurrence between the groups (combining all the tumors), but the study had insufficient power to detect an effect on any one malignancy.
Such disparate results highlight the difficulties facing investigators in this field. Differences in tumors, surgical techniques and skill, and patient populations, as well as challenges in defining recurrence and difficulty with long-term follow-up all hamper the ability to draw firm conclusions. Clinicians will need to wait for the results of ongoing clinical trials for more reliable information.
One interesting observation that can be made about these results is to note the direction of association. A completely unadjusted odds ratio for treated recurrence from our data is 1.15 (chi square P value = 0.24). In the unadjusted conditional logistic regression model, the odds ratio for recurrence is 1.38 (P = 0.06), strongly suggestive of an association. This discrepancy arises from the fact that the regression model includes clustering due to hospitals. Why might this be so? One plausible explanation is that patients at high-volume academic medical centers would be more likely to receive epidural analgesia and would also probably be more likely to have aggressive or unusual disease presentation. They may potentially also have been treated more aggressively after gastrectomy which would appear in this analysis as a recurrence event. Without more patient- and hospital-specific data, this remains hypothetical and is an avenue for further research. Of note, we observed a similar phenomenon in a previous study evaluating colorectal cancer recurrence that used a similar design.19
Of note, we found a significant association between perioperative blood transfusion and an increased mortality hazard. This is consistent with the recognized negative effects of blood transfusion. In colon cancer patients, transfusion has previously been associated with increased odds of recurrence after surgery.19,38 In a wider surgical population, Glance at al found an association between blood transfusion and an increase in multiple perioperative complications, including death.39 Given the observational nature of the present study, however, it is still unknown if transfusion causes worse outcomes or is simply a marker of a poor clinical course.
There are important limitations to this analysis. It suffers from the same weaknesses inherent to all observational studies, including susceptibility to bias and confounding. Because gastric cancer is relatively uncommon (and we had to exclude many patients to generate our analytic cohort), our study may be underpowered to detect a true difference between groups. Alternatively, our results may be explained by unmeasured confounders or selection bias. The post hocpower analysis would support this argument, but these analyses must be interpreted cautiously. We used propensity scores in an attempt to adjust for potential selection bias in treatment assignment, but we were limited to the available covariates. Because propensity scores only balance groups on known variables, unmeasured confounders may still remain.
Additionally, this database, because it is based on Medicare claims, contains limited clinical data. The timing of epidural use (preoperative, intraoperative, or only for a few days postoperatively) and the epidural failure rate are also unclear from this database. Without any other means of identifying epidural use in this database, some level of misclassification will be inevitable. The misclassification, however, should bias toward a null result. Additionally, the non-epidural group may have received a variety of analgesic modalities (eg, bolus opioids, intrathecal morphine, patient-controlled analgesia, local infiltration) which would increase the heterogeneity of the group.
Finally, the database does not have a variable indicating cancer recurrence. Therefore, we were not able to assess disease-free survival. Using subsequent treatments as proxies for recurrence may lead to cases of untreated recurrence going undetected, which could bias the result of the recurrence analysis. This is likely, as our detected rate of treated recurrence (25.6%) is on the lower end of the spectrum of reported recurrence rates. Also, due to the limitations of the database, we may have omitted clinically significant covariates from our statistical models and were unable to directly examine nonsurgical complications (such as cardiac events). Finally, our dataset was limited to individuals aged ≥ 66 years.
This study, however, has several strengths worth mentioning. The SEER-Medicare database is large, provides detailed tumor information, and is maintained by the National Cancer Institute. The data quality is very good and provides information from across the United States. This combination of clinical data with Medicare claims allows many questions to be addressed which would be impractical to study prospectively.
In conclusion, this population-based cohort study suggests that epidural anesthesia and/or analgesia is not associated with reduced treated recurrence or improved survival in patients with local or regional stage gastric cancer undergoing resection. Prospective studies are still needed.
Acknowledgments
Funding:
Case Western Reserve University/Cleveland Clinic CTSA UL1RR024989 (KC)
Case Western Reserve University GI SPORE P50 CA150964 (LC)
NIH K12 CA076917 (LC)
The authors thank Fang Xu, PhD, MS, for her work in the initial design of this statistical analysis.
This study used the linked SEER-Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the Applied Research Program, NCI; the Office of Research, Development and Information, CMS; Information Management Services (IMS) Inc; and the Surveillance, Epidemiology, and End Results (SEER) Program tumor registries in the creation of the SEER-Medicare database.
Footnotes
http://healthservices.cancer.gov/seermedicare/program/comorbidity.html, last accessed 10/10/2011
Departmental Attribution:
Anesthesiology Institute, Cleveland Clinic, Cleveland, OH
Prior Presentation:
Anesthesiology 2013, the annual meeting of the American Society of Anesthesiologists, October 12–16, 2013, San Francisco, CA
Conflict of Interest:
The authors declare no conflict of interest.
References
- 1.Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer J Clin. 2013;63:11–30. doi: 10.3322/caac.21166. [DOI] [PubMed] [Google Scholar]
- 2.Marrelli D, De Stefano A, de Manzoni G, Morgagni P, Di Leo A, Roviello F. Prediction of recurrence after radical surgery for gastric cancer: a scoring system obtained from a prospective multicenter study. Ann Surg. 2005;241:247–255. doi: 10.1097/01.sla.0000152019.14741.97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Marrelli D, Pedrazzani C, Morgagni P, et al. Changing clinical and pathological features of gastric cancer over time. Br J Surg. 2011;98:1273–1283. doi: 10.1002/bjs.7528. [DOI] [PubMed] [Google Scholar]
- 4.Eom BW, Yoon H, Ryu KW, et al. Predictors of timing and patterns of recurrence after curative resection for gastric cancer. Dig Surg. 2010;27:481–486. doi: 10.1159/000320691. [DOI] [PubMed] [Google Scholar]
- 5.Eschwège P, Dumas F, Blanchet P, et al. Haematogenous dissemination of prostatic epithelial cells during radical prostatectomy. Lancet. 1995;346:1528–1530. doi: 10.1016/s0140-6736(95)92054-4. [DOI] [PubMed] [Google Scholar]
- 6.Denis MG, Lipart C, Leborgne J, et al. Detection of disseminated tumor cells in peripheral blood of colorectal cancer patients. Int J Cancer. 1997;74:540–544. doi: 10.1002/(sici)1097-0215(19971021)74:5<540::aid-ijc11>3.0.co;2-a. [DOI] [PubMed] [Google Scholar]
- 7.Holmgren L, O’Reilly MS, Folkman J. Dormancy of micrometastases: balanced proliferation and apoptosis in the presence of angiogenesis suppression. Nat Med. 1995;1:149–153. doi: 10.1038/nm0295-149. [DOI] [PubMed] [Google Scholar]
- 8.Bar-Yosef S, Melamed R, Page GG, Shakhar G, Shakhar K, Ben-Eliyahu S. Attenuation of the tumor-promoting effect of surgery by spinal blockade in rats. Anesthesiology. 2001;94:1066–1073. doi: 10.1097/00000542-200106000-00022. [DOI] [PubMed] [Google Scholar]
- 9.Ben-Eliyahu S, Page GG, Yirmiya R, Shakhar G. Evidence that stress and surgical interventions promote tumor development by suppressing natural killer cell activity. Int J Cancer. 1999;80:880–888. doi: 10.1002/(sici)1097-0215(19990315)80:6<880::aid-ijc14>3.0.co;2-y. [DOI] [PubMed] [Google Scholar]
- 10.Page GG, Blakely WP, Ben-Eliyahu S. Evidence that postoperative pain is a mediator of the tumor-promoting effects of surgery in rats. Pain. 2001;90:191–199. doi: 10.1016/s0304-3959(00)00403-6. [DOI] [PubMed] [Google Scholar]
- 11.Lutgendorf SK, Cole S, Costanzo E, et al. Stress-related mediators stimulate vascular endothelial growth factor secretion by two ovarian cancer cell lines. Clin Cancer Res. 2003;9:4514–4521. [PubMed] [Google Scholar]
- 12.Brand JM, Kirchner H, Poppe C, Schmucker P. The effects of general anesthesia on human peripheral immune cell distribution and cytokine production. Clin Immunol Immunopathol. 1997;83:190–194. doi: 10.1006/clin.1997.4351. [DOI] [PubMed] [Google Scholar]
- 13.Shapiro J, Jersky J, Katzav S, Feldman M, Segal S. Anesthetic drugs accelerate the progression of postoperative metastases of mouse tumors. J Clin Invest. 1981;68:678–685. doi: 10.1172/JCI110303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Beilin B, Shavit Y, Hart J, et al. Effects of anesthesia based on large versus small doses of fentanyl on natural killer cell cytotoxicity in the perioperative period. Anesth Analg. 1996;82:492–497. doi: 10.1097/00000539-199603000-00011. [DOI] [PubMed] [Google Scholar]
- 15.Yeager MP, Colacchio TA, Yu CT, et al. Morphine inhibits spontaneous and cytokine-enhanced natural killer cell cytotoxicity in volunteers. Anesthesiology. 1995;83:500–508. doi: 10.1097/00000542-199509000-00008. [DOI] [PubMed] [Google Scholar]
- 16.O’Riain SC, Buggy DJ, Kerin MJ, Watson RW, Moriarty DC. Inhibition of the stress response to breast cancer surgery by regional anesthesia and analgesia does not affect vascular endothelial growth factor and prostaglandin E2. Anesth Analg. 2005;100:244–249. doi: 10.1213/01.ANE.0000143336.37946.7D. [DOI] [PubMed] [Google Scholar]
- 17.Exadaktylos AK, Buggy DJ, Moriarty DC, Mascha E, Sessler DI. Can anesthetic technique for primary breast cancer surgery affect recurrence or metastasis? Anesthesiology. 2006;4:660–664. doi: 10.1097/00000542-200610000-00008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Biki B, Mascha E, Moriarty DC, Fitzpatrick JM, Sessler DI, Buggy DJ. Anesthetic technique for radical prostatectomy surgery affects cancer recurrence: a retrospective analysis. Anesthesiology. 2008;109:180–187. doi: 10.1097/ALN.0b013e31817f5b73. [DOI] [PubMed] [Google Scholar]
- 19.Cummings KC, 3rd, Xu F, Cummings LC, Cooper GS. A comparison of epidural analgesia and traditional pain management effects on survival and cancer recurrence after colectomy: apopulation-based study. Anesthesiology. 2012;116:797–806. doi: 10.1097/ALN.0b013e31824674f6. [DOI] [PubMed] [Google Scholar]
- 20.Gottschalk A, Ford JG, Regelin CC, et al. Association between epidural analgesia and cancer recurrence after colorectal cancer surgery. Anesthesiology. 2010;113:27–34. doi: 10.1097/ALN.0b013e3181de6d0d. [DOI] [PubMed] [Google Scholar]
- 21.Myles PS, Peyton P, Silbert B, et al. Perioperative epidural analgesia for major abdominal surgery for cancer and recurrence-free survival: randomised trial. BMJ. 2011:342. doi: 10.1136/bmj.d1491. [DOI] [PubMed] [Google Scholar]
- 22.Rodgers A, Walker N, Schug S, et al. Reduction of postoperative mortality and morbidity with epidural or spinal anaesthesia: Results from overview of randomised trials. BMJ. 2000;321:1493. doi: 10.1136/bmj.321.7275.1493. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Christopherson R, James KE, Tableman M, Marshall P, Johnson FE. Long-term survival after colon cancer surgery: A variation associated with choice of anesthesia. Anesth Analg. 2008;107:325–332. doi: 10.1213/ane.0b013e3181770f55. [DOI] [PubMed] [Google Scholar]
- 24.Gupta A, Bjornsson A, Fredriksson M, Hallbook O, Eintrei C. Reduction in mortality after epidural anaesthesia and analgesia in patients undergoing rectal but not colonic cancer surgery: A retrospective analysis of data from 655 patients in Central Sweden. Br J Anaesth. 2011;107:164–170. doi: 10.1093/bja/aer100. [DOI] [PubMed] [Google Scholar]
- 25.Silber JH. Medicare claims and anesthesia clinical research: A perfect match. Reg Anesth Pain Med. 2003;28:259–261. doi: 10.1016/s1098-7339(03)00195-0. [DOI] [PubMed] [Google Scholar]
- 26.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):IV-3–18. doi: 10.1097/01.MLR.0000020942.47004.03. [DOI] [PubMed] [Google Scholar]
- 27.Diggs JC, Xu F, Diaz M, Cooper GS, Koroukian SM. Failure to screen: Predictors and burden of emergency colorectal cancer resection. Am J ManagCare. 2007;13:157–164. [PubMed] [Google Scholar]
- 28.Wu CL, Anderson GF, Herbert R, Lietman SA, Fleisher LA. Effect of postoperative epidural analgesia on morbidity and mortality after total hip replacement surgery in Medicare patients. Reg Anesth Pain Med. 2003;28:271–278. doi: 10.1016/s1098-7339(03)00095-6. [DOI] [PubMed] [Google Scholar]
- 29.Silber JH, Kennedy SK, Even-Shoshan O, Chen W, Koziol LF, Showan AM, Longnecker DE. Anesthesiologist direction and patient outcomes. Anesthesiology. 2000;93:152–63. doi: 10.1097/00000542-200007000-00026. [DOI] [PubMed] [Google Scholar]
- 30.Silber JH, Kennedy SK, Even-Shoshan O, Chen W, Mosher RE, Showan AM, Longnecker DE. Anesthesiologist board certification and patient outcomes. Anesthesiology. 2002;96:1044–1052. doi: 10.1097/00000542-200205000-00004. [DOI] [PubMed] [Google Scholar]
- 31.Earle CC, Nattinger AB, Potosky AL, et al. Identifying cancer relapse using SEER-Medicare data. Med Care. 2002;40(8 Suppl):IV-75–81. doi: 10.1097/00005650-200208001-00011. [DOI] [PubMed] [Google Scholar]
- 32.Knopf KB, Warren JL, Feuer EJ, Brown ML. Bowel surveillance patterns after a diagnosis of colorectal cancer in Medicare beneficiaries. Gastrointes Endosc. 2001;54:563–571. doi: 10.1067/mge.2001.118949. [DOI] [PubMed] [Google Scholar]
- 33.Warren JL, Harlan LC, Fahey A, et al. Utility of the SEER-Medicare data to identify chemotherapy use. Med Care. 2002;40:IV-55–61. doi: 10.1097/01.MLR.0000020944.17670.D7. [DOI] [PubMed] [Google Scholar]
- 34.Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613–619. doi: 10.1016/0895-4356(92)90133-8. [DOI] [PubMed] [Google Scholar]
- 35.Klabunde CN, Potosky AL, Legler JM, Warren JL. Development of a comorbidity index using physician claims data. J Clin Epidemiol. 2000;53:1258–1267. doi: 10.1016/s0895-4356(00)00256-0. [DOI] [PubMed] [Google Scholar]
- 36.Cooper GS, Xu F, Barnholtz Sloan JS, Koroukian SM, Schluchter MD. Management of malignant colonic polyps: A population-based analysis of colonoscopic polypectomy versus surgery. Cancer. 2012;118:651–659. doi: 10.1002/cncr.26340. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Wei LJ, Lin DY, Weissfeld L. Regression analysis of multivariate incomplete failure time data by modeling marginal distributions. JASA. 1989;84:1063–1073. [Google Scholar]
- 38.Amato A, Pescatori M. Perioperative blood transfusions for the recurrence of colorectal cancer. Cochrane Database Syst Rev. 2006:CD005033. doi: 10.1002/14651858.CD005033.pub2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Glance LG, Dick AW, Mukamel DB, et al. Association between intraoperative blood transfusion and mortality and morbidity in patients undergoing noncardiac surgery. Anesthesiology. 2011;114:283–292. doi: 10.1097/ALN.0b013e3182054d06. [DOI] [PubMed] [Google Scholar]