| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Surgery, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, California; and the Department of Surgery, West Los Angeles Veterans Affairs Hospital, Los Angeles, California.
Correspondence: Address correspondence and reprint requests to: Jessica B. O’Connell, MD, Department of Surgery, David Geffen School of Medicine, University of California, Los Angeles, 10833 Le Conte Avenue, Room 72-215, Los Angeles, CA 90095; Fax: 818-501-4017; E-mail: jbocjboc{at}hotmail.com
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods: By using the Surveillance, Epidemiology, and End Results database (1988–1997), patients (
40 years) were identified with localized adenocarcinoma of the breast, esophagus, stomach, pancreas, colon, or rectum; non–small-cell lung carcinoma; and sarcoma (n = 200,360). Rates of cancer-directed surgery (CDS) were compared across age groups (at 5-year intervals). Multivariate regression was used to identify predictors of receipt of CDS in each tumor group.
Results: Rates of CDS declined steadily with increasing age for all nine localized tumors. Most striking were the low rates of CDS for patients >70 years with esophagus, stomach, pancreas, and lung cancers (range, 0%–83%). However, CDS rates were >90% for breast and colon and >84% for rectal cancer in all age groups. Multivariate regression found lower odds of CDS for elderly patients for all cancers except colon. For example, age significantly decreased the odds of receiving CDS beginning at 60 years for lung cancer (odds ratio [OR], .550; P = .03), at 70 years for liver cancer (OR, .109; P = .003), and at 80 years for pancreatic cancer (OR, .120; P < .05).
Conclusions: Although CDS for localized disease is being performed regularly in the elderly for some cancers (e.g. breast, colon, and rectum), this analysis shows that elderly patients are not receiving surgery for many potentially curable cancers. Whether these rates are appropriate or too low requires further evaluation. This is particularly essential because our population is aging.
Key Words: Elderly • Cancer • Surgery • Cancer registry • SEER • Quality
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Prior studies have reported on the underuse of radiation and chemotherapy for cancer treatment in the elderly.3–5 For example, one study found that only 34% of colon cancer patients 80 to 84 years old and 11% of those 85 to 89 years old received appropriate (according to the 1990 National Institutes of Health Consensus Conference guidelines) postoperative chemotherapy.3 However, few studies have similarly examined the rates of use (or underuse) of surgical cancer resection with regard to the elderly population. To address this issue further, this study used a nationwide, population-based data sample and analyzed the use of surgery in patients older than 40 years of age with local-stage disease (i.e., tumors that should be amenable to resection) for nine common surgically treated cancers.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
SEER collects patient records from multiple sites across the United States and is regarded as a model population-based tumor registry. This national program includes 12 regional registries, which comprise approximately 14% of the population. The database was designed to reflect the overall characteristics of the United States, including the variety of racial/ethnic groups, geographical locations, and types of cities.6
The cancer types were selected by single SEER site codes as follows: breast, esophagus, stomach, pancreas, and lung. "Liver" tumors included site codes for liver and intrahepatic bile duct tumors. "Colon" location was defined as cecum, ascending colon, hepatic flexure, transverse colon, splenic flexure, descending colon, sigmoid colon, and large intestine not otherwise specified. "Rectal" cancer location was defined by the rectosigmoid junction and rectum. "Sarcoma" cancers included soft tissue, bone and joint, and retroperitoneum tumors. Appendiceal and in situ tumors were excluded.
The tumors were further limited by specific histologies as defined by individual International Classification of Diseases for Oncology, 2nd edition, codes: adenocarcinomas (8020–8022, 8140–8145, 8200, 8201, 8210, 8211, 8220, 8221, 8230, 8231, 8260–8263, 8470–8473, 8480, 8481, and 8490) and additionally for breast adenocarcinoma (8010, 8033, 8050, 8051, 8500–8504, 8520–8522, 8530, and 8540–8543), hepatocellular carcinoma (8170, 8171, and 8180), non–small-cell lung carcinoma (8000–8004, 8020–8022, 8030–8032, 8050, 8070–8076, 8140–8145, 8200, 8201, 8210, 8211, 8220, 8221, 8230, 8231, 8250, 8251, 8260–8263, 8323, 8470–8473, 8480, 8481, 8490, 8560), and sarcoma (8800–8804, 8810, 8811, 8832, 8840, 8850–8858, 8890, 8895, 8896, 8900, 8901, 8990, 9040–9044, 9120, 9130, 9170, 9180–9184, 9190, 9220, 9221, 9230, 9231, 9240, and 9260).
Local-stage disease was initially determined by the "SEER historic stage" and was further restricted by the "extent of disease" coding in SEER to include only tumors that were confined to the organ of origin and not invading any adjacent tissues or structures. Furthermore, for liver tumors, any tumor >5 cm was excluded. In this manner, we attempted to select only the tumors that would be amenable to surgical resection.
Patient Demographics
Demographic information recorded for each patient included age (organized into 11 age categories: 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70–74, 75–79, 80–84, 85–89, and 90 years), sex, race/ethnicity (white, black, Hispanic, Asian, and other), marital status (single vs. married at diagnosis), birthplace (born inside vs. outside of the United States), and SEER registry location (Alaska, Atlanta, Connecticut, Detroit, Hawaii, Iowa, Los Angeles, New Mexico, San Francisco, San Jose, Seattle, or Utah). These data were also included in multivariate analyses as covariates.
Tumor Characteristics
Cancer-specific data evaluated for each patient included tumor location, stage at presentation, specific histology, and tumor size. Tumor stage was evaluated by using the SEER historic staging system (localized, regional, and distant). Only localized tumors were included in the analysis. Tumor size was measured in centimeters.
Surgical Treatment and Reasons for Not Undergoing Surgery
Whether patients underwent cancer-directed surgery (CDS; i.e. surgery intended for cure) was noted, and the reason for not undergoing CDS was stratified by "not recommended," "contraindicated," "refused," and "unknown." This variable in SEER is collected directly from the patient’s medical record. "Not recommended" is coded if the physician believed that surgery was not the best treatment option, "contraindicated" is coded if a medical contraindication exists, and "refused" is coded if the patient refused surgery.6
Statistical Analysis and Survival Analysis
Statistical comparisons between groups were completed with the 
2 test of proportions or a two-sided Fisher’s exact test when expected values were <5. P values <.05 were considered statistically significant.
Multivariate logistic regression analysis for each cancer type was performed to predict the receipt of CDS. Independent covariates were age group, sex, race/ethnicity, marital status at diagnosis, birthplace, and tumor size. Another multivariate regression analysis was performed to predict surgery being "not recommended" as the reason for not undergoing CDS. Independent covariates for this separate analysis were tumor type, age group, sex, race/ethnicity, marital status at diagnosis, birthplace, SEER registry, and tumor size.
Survival analysis was also performed with a multivariate Cox proportional hazards model. For this analysis, independent covariates included for analysis were age group, sex, race/ethnicity, marital status at diagnosis, birthplace, tumor size, registry, and reason for not undergoing CDS. All statistical analyses were completed with SAS version 8.02 (SAS Institute, Cary, NC).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
CDS Treatment
Overall rates of CDS ranged from 26.2% (pancreas) to 99.1% (breast; Fig. 2). Evaluation of the 11 age groups revealed that rates of CDS declined steadily with increasing age for all 9 tumor types. Most striking were the low rates of CDS for patients older than 70 years with lung, esophagus, stomach, liver, and pancreas cancers. Figure 3 graphically demonstrates that elderly patients received significantly less CDS compared with younger patients. Esophageal cancer patients older than 70 years underwent CDS 54.0% of the time, compared with those younger than 55 years, who had CDS more than 75.0% of the time. Similarly, for lung tumors, patients younger than 70 years received CDS more than 75.0% of the time, compared with only 40.0% of patients over the age of 80. However, rates were >90.0% for all age groups with colon cancer, >84.0% for patients with rectal cancer, and >95.0% for breast cancer patients, even those >90 years old.
|
|
Race/ethnicity contributed to differences in rates of CDS for six of the nine tumors (i.e., breast, lung, liver, pancreas, colon, and rectum). The largest differences were seen between whites and blacks. For example, 68.0% of white patients with lung cancer received CDS, compared with 58.0% of blacks (P < .0001). Similar results were found for liver cancer (49.0% CDS in whites vs. 23.0% in blacks; P = .004) and pancreas cancer (27.0% in whites vs. 16.0% blacks; P = .003). Racial/ethnic differences were also seen for Hispanics and Asians, with 68.0% of white lung cancer patients receiving CDS compared with 60.2% of Hispanics (P < .0001 vs. whites) and 63.9% of Asians (P = .003 vs. whites). It is interesting to note that Asians received significantly more CDS for liver cancer (63.7% vs. 49.0% for whites; P = .03) and for stomach cancer (86.5% vs. 74.7% for whites; P < .0001).
Tumor size was a factor for several cancers when comparing patients who received CDS with those who did not receive surgery. Patients with breast cancer who underwent CDS had a mean tumor size of 1.7 cm, compared with 3.1 cm in patients who did not have surgery (P < .0001). Similarly, statistically significant tumor size differences were seen for lung (3.0 cm for those who had CDS vs. 3.5 cm for those who did not have surgery; P < .0001), esophagus (3.8 cm for CDS vs. 6.5 cm for no surgery; P < .0001), and liver (3.0 cm for CDS vs. 3.6 cm for no surgery; P < .0001) tumors. Tumor size differences (for surgery vs. no surgery) were not seen for patients with stomach, pancreas, colon, or rectum tumors or sarcomas.
Reason for Not Undergoing CDS
For the entire cohort, the rate of CDS was 93.0%. Surgery was "not recommended" for 3.4% of patients, "contraindicated" for 1.1%, and "refused" by .7%; the reason was unknown for 1.8% of patients (Table 2).
|
Multivariate Regression Analysis and Cox Proportional Survival Analysis
Multivariate regression analysis found a decline in the odds of receiving CDS for elderly patients for all cancers except colon (adjusted for age, race/ethnicity, sex, marital status, birthplace, and tumor size). Age significantly decreased the odds of receiving CDS beginning at age 60 years for lung tumors (odds ratio [OR], .550; P = .03; Table 3); at 70 years for liver tumors (OR, .109; P = .003); at 80 years for breast, esophagus, and pancreas tumors (OR, .440, .126, and .120, respectively; P < .05 for all); at 85 years for stomach tumors and sarcoma (OR, .077 and .114, respectively; P < .05 for both); and at age 90 for rectal cancer (OR, .256; P = .04; Tables 4 and 5 
).
|
|
|
90 years: OR, 7.518; 95% CI, 4.624–12.222; P < .01; Table 6).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This study is one of the first to focus on the rates of CDS, by age, for a variety of local-stage cancers (i.e., cancers that should ideally be resected for any patient). We found that variations in the use of CDS for elderly patients exist for many cancers. Whereas colorectal and breast cancer patients have excellent rates of receiving CDS (>90% for colon, >84% for rectum, and >95% for all patients with breast cancer), other tumors—including lung, esophagus, stomach, liver, and pancreas—were found to have lower rates. As stated previously, these findings are probably due to multiple factors, including the difficulty, length, and morbidity of the procedures (e.g., mastectomy versus esophagectomy); the overall comorbidities of the patient (e.g., congestive heart failure and diabetes); and the specific cancer-related comorbidities associated with the development of the particular cancer (e.g., comorbidities associated with long-time smoking in lung cancer, such as chronic obstructive pulmonary disease and emphysema). Regardless of the role of the comorbidities, however, studies have noted that the main causes of postoperative mortalities are cardiopulmonary, and if preoperative and postoperative care can be focused on preventing these complications, better-quality care for the elderly population could be attained.11
A notable finding of this study involves the age cutoffs that were identified in multivariate analysis predicting the receipt of CDS. As age increased for all the tumors studied (except colon), there was a distinct age cutoff for each of the tumors at which there were statistically significantly decreased odds of receiving CDS. As one might expect, the age cutoffs substantially varied by tumor (Tables 4 and 5
); consequently, the proportion of patients who did not receive CDS varied as well, ranging from low (2.4% for those >90 years old with rectal cancer, 5.5% for sarcoma patients >85 years old, and 12.7% for breast cancer patients >80 years old) to much higher (37.0% for liver cancer patients >70 years and 84.0% for lung cancer patients >60 years old; Table 5). This variation is likely affected by the presence and severity of the comorbidities associated with the various cancers (e.g., lung cancer and chronic obstructive pulmonary disease; liver cancer and cirrhosis).
Although it has been demonstrated that elderly patients receive less CDS than younger patients, it remains unclear whether these rates are appropriate. In other words, do factors including the level of comorbidity and the severity of the operative procedure adequately explain the varying rates of CDS in the elderly? This issue is difficult to address, particularly in a secondary data analysis.
If we examine the literature on other areas of variation, however, we see that "varying rates" often equate to different levels of care. For example, studies have shown that there are varying rates of sphincter-sparing procedures for rectal cancer (as opposed to abdominoperineal resection). More specifically, significantly more sphincter-sparing procedures are performed in teaching hospitals and specialty centers and by specialist surgeons.12,13 In all likelihood, the expertise or knowledge required to perform sphincter-sparing operations is not equivalent across all facilities or providers, and this leads to variation. Although unproven in our study, it is not unreasonable to surmise that a similar situation could exist for treating the elderly—i.e., that certain facilities or providers offer and provide better care for this segment of the population. Further work is needed in this regard.
As such, determining whether the rates of CDS are appropriate is important concerning the potential implications of this study. If these rates of CDS are inappropriate, then specific guidelines may need to be developed regarding the indications of (or contraindications for) surgery, particularly in the elderly. There is a growing literature regarding the preoperative evaluation of surgical patients, and it might be useful to fold this literature into the treatment of specific diseases (e.g., specific cancers). We foresee that guidelines would probably focus more on functional status than on age alone. These potential guidelines would be important not only for defining the appropriateness of cancer surgery in the elderly, but also for improving the quality of surgery overall.
Although this is the largest population-based study to date evaluating surgery for surgical cancers in the elderly, there are potential limitations with the results. First, although the SEER registry maintains stringent quality-control measures to prevent coding errors, possible miscoding and inaccurate data may be present. SEER does, however, uphold several measures to ensure accuracy and maintains the highest level of certification of data quality and completeness, as reported by the North American Association of Central Cancer Registries.14 Second, the registry lacks some clinically relevant treatment data. One of the most important issues in this regard is information pertaining to the presence and severity of patient comorbidities. As such, evaluation of completely risk-adjusted rates of CDS was not possible. Finally, it is possible that not all tumors evaluated in this study were truly localized. We did tailor our patient selection to include only patients with resectable disease (on the basis of several clinicopathologic factors available in SEER, including size and the pathologic or radiographical extent of disease), but if a patient did not undergo surgery at all (and therefore did not have pathologic information), we do not know that the patient had localized disease. This could potentially add patients to our cohort who actually had higher-stage disease, possibly skewing the results.
Despite these possible limitations, there are many advantages to the population-based SEER registry. First, we were able to analyze 200,360 patients with 9 surgically treated cancers over 10 years—this is much larger than any other study. Second, SEER is nationally representative because it is a cross-sectional sample of approximately 14% of the US population that is ethnically and socioeconomically diverse.6 As such, these findings should generalize to the overall population and provide the opportunity to examine and discuss issues at a national health policy level.
In summary, we found, by using a national population-based data sample, that elderly patients with lung, esophagus, stomach, liver, and pancreatic cancers do not receive CDS for potentially curable localized disease. Surgical intervention’s being "not recommended" was the most common reason. This is in contrast to >90% of similarly staged patients of all ages with breast and colon cancer and >84% of rectal cancer patients receiving appropriate surgery. Given the rapidly aging population, further study should focus on the specific reasons why these patients are not receiving surgery, on working toward determining whether these rates are appropriate, and on development of guidelines for cancer surgery, specifically in the elderly.
|
FOOTNOTES |
|---|
This study used the Surveillance, Epidemiology, and End Results national cancer database to evaluate rates of cancer-directed surgery for patients with local-stage cancers. For some cancers, the elderly do not receive cancer-directed surgery as often as younger patients for these potentially curable tumors.
Received for publication March 19, 2004. Accepted for publication August 1, 2004.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. C. C. Tan, N. G. Coburn, N. N. Baxter, A. Kiss, and C. H. L. Law Surgical Management of Intrahepatic Cholangiocarcinoma - A Population-Based Study Ann. Surg. Oncol., February 1, 2008; 15(2): 600 - 608. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. D. Palnum, P. Petersen, H. T. Sorensen, A. Ingeman, J. Mainz, P. Bartels, and S. P. Johnsen Older patients with acute stroke in Denmark: quality of care and short-term mortality. A nationwide follow-up study Age Ageing, January 1, 2008; 37(1): 90 - 95. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. R. Lange and C. M. Balch Melanoma in Older Patients ASCO Educational Book, January 1, 2008; 2008(1): 399 - 402. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Taneja, S. Mandayam, Z. Z. Kayani, Y.-F. Kuo, and V. B. Shahinian Comparison of Stage at Diagnosis of Cancer in Patients Who Are on Dialysis versus the General Population Clin. J. Am. Soc. Nephrol., September 1, 2007; 2(5): 1008 - 1013. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. N. Baxter, B. A. Whitson, and T. M. Tuttle Trends in the Treatment and Outcome of Pancreatic Cancer in the United States Ann. Surg. Oncol., April 1, 2007; 14(4): 1320 - 1326. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Caywood, R. J. Gray, J. Hentz, and B. A. Pockaj Older Age Independently Predicts a Lower Risk of Sentinel Lymph Node Metastasis in Breast Cancer Ann. Surg. Oncol., December 1, 2005; 12(12): 1061 - 1065. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
