| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Departments of Pathology (RVR) and Surgery (LD, BYS, SMW), Stanford University School of Medicine (ACB), Stanford University, Palo Alto, California; and the Palo Alto Veterans Affairs Health Care System, Palo Alto, California (RVR, LD, BYS, SMW).
Correspondence: Address correspondence and reprint requests to: Sherry M. Wren, MD, FACS, Stanford University Department of Surgery, G112 PAVAHCS, 3801 Miranda Avenue, Palo Alto, CA 94304; Fax: 650-852-3430; E-mail: swren{at}stanford.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Methods: A retrospective review, from 1991 to 2002, of our institution’s pathology and cancer registry databases for rectal cancers. CRCS data were obtained from the Veterans Administration Prevention Disease Index. Cancer stage at diagnosis was compared before and after the directive and was compared with the National Cancer Data Base and the Surveillance, Epidemiology, and End Results data.
Results: The rate of CRCS was 55% in 1998 and increased to 75% in 2003. During the 11 years studied, a total of 147 rectal cancers were diagnosed. After the Veterans Administration directive, there was a significant increase in stage 0 cancers (P < .02) and an overall migration to earlier-stage cancers. Our Veterans Administration hospital had a significantly greater percentage of stage 0 cancers both before (P < .007) and after the directive (P < .00) and had fewer stage 3 cancers after the directive (P < .03) compared with National Cancer Data Base data. Compared with Surveillance, Epidemiology, and End Results data, the Palo Alto Veterans Affairs Health Care System had more local disease (P < .03) and less regional disease (P < .006).
Conclusions: These data suggest that a monitored institutional directive may significantly increase early detection of rectal cancers. This should result in a greater survival rate and lower local recurrence rate, because survival is predicated on stage at presentation. This may serve as a model for other health-care systems.
Key Words: Colorectal cancer • Rectal cancer • Screening • Cancer staging
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Unfortunately, despite strong evidence of the mortality benefits, cost savings, and low morbidity of screening, rates in the United States remain lower than desired and are below screening rates for other cancers.20–22 Improvement in screening compliance should lead to detection of earlier-stage CRC. Low income, lack of health insurance, and lack of a usual source of care negatively affect the rate of screening.21,22 The Veterans Administration (VA) health-care system provides access to a broad range of health-care services for veterans at no or low cost according to their ability to pay. As an integrated health-care system providing services to a captured population of patients, it is an ideal environment in which broad-reaching screening programs can be implemented and potential benefits of the screening programs can be monitored and realized.
The VA health-care system chose CRCS as one of its performance measures in a 1998 initiative to improve care. Since then, an independent, external review organization has measured CRCS rates in primary-care patients. These data are part of a facility’s overall performance evaluation and are compared across VA hospitals nationwide. The Palo Alto VA Health Care System (PAVAHCS) is a large regional primary- and tertiary-care provider. There are approximately 40,000 patients enrolled in our primary-care program aged more than 50 years and from a large geographical area.
At the PAVAHCS, we investigated the effect of the VA screening directive on CRCS rates and on the stage at presentation of rectal cancers. Although CRCS targets disease of the entire colon and rectum, we chose to examine rectal cancer specifically because of its treatment differences compared with colon cancer and the effect of rectal cancer treatment on our patients and the health-care system. Rectal cancers that are locally advanced (T3) or have regional nodal involvement (N1–2) require a treatment regimen of chemotherapy and radiotherapy. Radiotherapy can be a challenge to arrange in a convenient fashion for our patients and their families. Lower-stage cancers are clearly better for the patient and require no other adjuvant treatments. In addition, we compared our data with those of two nationally recognized cancer databases as benchmarks of cancer stages in California and the nation as a whole.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
In addition, cancers found through the pathology database were cross-verified with the PAVAHCS’s cancer registry database. The cancer registry database was searched by using the codes for rectum (C20.9) and rectosigmoid (C19.9), and only cancers <20 cm from the anal verge were included. If any discrepancy was found between databases, a full record review was performed to verify the correct stage at first presentation. Finally, a full record review was performed by the authors (S.M.W., A.C.B., and L.D.). Because the VA screening initiative occurred in 1998, rectal cancers were grouped into a preinitiative time period (1991–1998) and a postinitiative time period (1999–2002).
PAVAHCS Screening Rates
CRCS rates for PAVAHCS are published in an internal publication of the VA Preventive Health Initiative.23 The eligible populations were defined by the Veterans Health Care Administration as patients in the primary clinic cohort who were at least 52 years old at the time of the qualifying visit and who did not have a terminal condition. Screening rates were determined by using an External Peer Review Program (EPRP) under contract to an outside organization. A random sample of at least 1% of all primary-care charts for enrolled patients were reviewed yearly. A target rate is set by the Veterans Health Care Administration, and each facility’s performance is monitored against this goal and compared directly among other VA facilities.
Successful CRCS was defined by the Veterans Health Care Administration as three FOBT cards or one positive card within 12 months, flexible sigmoidoscopy in the last 5 years, or colonoscopy at a 10-year interval. The specific screening test used was not noted in the EPRP documents. Tests that were completed in the private sector or at another VA medical center were counted as long as the documentation was adequate to ascertain that the test had been accomplished in the accepted time window. Full details of the screening definitions and requirements can be obtained from the Office of Quality and Performance, Veterans Health Administration, Department of Veterans Affairs.24
National Benchmark Data
National and state-specific data on rectal cancer stage at presentation were obtained for comparison from the National Cancer Data Base (NCDB) and Surveillance, Epidemiology, and End Results (SEER) benchmark reports.25,26 The searches were constructed to retrieve only data on male patients. Cases were selected from the SEER Cancer Statistics Review for the years 1992 to 1999 (all available), "all regions," and cancers of the "rectum." Because the SEER database reports cancers only in the categories of local, regional, and distant, we placed our stages into those groupings for the analysis. Cancer stages were grouped as follows; stage 0, 1, and 2 cancers were local, stage 3 cancers were regional, and stage 4 cancers were distant. NCDB national data were selected for the years 1998 and 2001, "all states," and "rectum." NCDB California state data were selected for the same years. Data from 1998 were used to compare with PAVAHCS preinitiative cancers (1991–1998). Data from 2001, the latest available year, were used to compare with PAVAHCS postinitiative data.
Statistical Analysis
We calculated the frequency distribution of rectal cancer stages and compared the PAVAHCS preinitiative and postinitiative data by using Fisher’s exact test. PAVAHCS data were compared with data from the two national databases by using this same test. Analysis of variance was used to evaluate demographic data. Significance was accepted at P < .05.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Of the rectal cancers found in this 11-year period, 35 (23%) were stage 0, 30 (20%) were stage 1, 26 (18%) were stage 2, 23 (16%) were stage 3, 29 (20%) were stage 4, and 4 (3%) were not able to be staged with the information available. Curative endoscopic polypectomy was performed in 28 of the stage 0 cancers. The VA directive on screening was implemented in 1998. Therefore, the data were organized into preinitiative (1991–1998) and postinitiative (1999–2002) groups for data analysis. There were 90 rectal cancers in the preinitiative group and 57 in the postinitiative group.
PAVAHCS EPRP Data
Before 1995, the rate of CRCS in our institution was <30% in random sample charts. In 1998, the year of the screening directive implementation, the rate was 55%. Since the directive’s implementation, the rate has continued to increase to 75% in 2003 (Table 1).23 VA-mandated screening targets increased from 57% in 1999 to 2001 to 70% in 2003. In 1999, the recorded rate for screening was artificially increased (80%) because the definition of acceptable screening required only that FOBT cards be distributed but did not require proof of results from the cards to be complete.24 Starting in 2000, screening was considered accomplished if results of the tests were documented within the specified time periods.
|
|
The overall distribution of stages within the NCDB did not appreciably change from 1998 to 2001. In distinct contrast, the distribution of stages at PAVAHCS dramatically changed in the preinitiative (1992–1998) and postinitiative (1999–2002) time periods (Fig. 2).
|
In the postinitiative time period, the greater percentage of stage 0 cancers (35%) at the PAVAHCS was even more pronounced compared with NCDB national (7%; P < .001) and California (6%; P < .001) data. Of great clinical importance, there was a significant decrease in lymph node–positive stage 3 cancers (9%) compared with NCDB national (20%; P < .03) and California (23%; P < .02) data. There was still a greater percentage of stage 4 cancers at the PAVAHCS (18%) compared with NCDB national (13%) and California (13%) data, but this difference was no longer statistically significant (Fig. 3).
|
PAVAHCS had a significantly greater percentage of patients with localized disease in both the preinitiative (54%; P < .03) and postinitiative (74%; P < .002) periods. In addition, a lower percentage of patients with regional disease was seen in the preinitiative (21%; P > .006) and postinitiative (9%; P < .001) periods. The percentage of patients with distant disease was greater at the PAVAHCS in both time periods, but this did not reach statistical significance (Fig. 4).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Our study considered all rectal cancers in one VA hospital over 11 years. During this time, CRCS became an official, national VA administrative mandate. Compliance with the mandate is monitored nationwide. As might be expected, as screening compliance increased, the stage at presentation of rectal cancers improved. There was a clear trend toward earlier-stage cancers at diagnosis after the 1998 screening directive. Stage 0 cancers in our facility jumped from 17% to 35% of all cancers diagnosed. There was also an important decrease in lymph node–positive stage 3 cancers, from 20% to 9%. The stage at presentation of rectal cancer dictates future treatment options and survival rates. Therefore, the observed increase in stage 0 and decrease in stage 3 cancers are especially relevant. Clearly, having fewer patients with regional nodal disease is an advantage, given the negative implications for survival and treatment strategies for rectal cancers in this group.
Compared with national and state data, the PAVAHCS had a greater percentage of early-stage rectal cancers in both time periods analyzed. Because there has been a slow improvement nationwide in CRCS and stage at presentation of CRC, we chose NCDB 1998 data to compare with our preinitiative years (1991–1998). Similarly, for postinitiative (1999–2002) comparisons, we chose 2001, the most recent benchmark available. SEER data are available only from 1992 to 1999. Thus, any possible benefits from improved nationwide screening may not be represented.
A higher rate of screening in the studied population is a possible explanation for the greater percentage of early cancers found at PAVAHCS. The PAVAHCS has had CRCS rates that are 20% to 30% higher than those found in studies of the general population since 1999, with screening consistently >70%.23 The Centers for Disease Control and Prevention analyzed data from the 1997, 1999, and 2001 Behavioral Risk Factor Surveillance System surveys. For persons aged 
50 years, screening rates have slowly increased from 41% in 1997 to 53.1% in 2001.20 A similar study using 1987, 1992, and 1998 National Health Interview Survey (NHIS) data found CRCS rates <40%.22 Although our facility’s CRCS rates may be overestimated because the distinction between screening and diagnostic or therapeutic endoscopies was not made, this is also true for both the NHIS and the Behavioral Risk Factor Surveillance System–based studies.
The VA may achieve greater screening success because it can eliminate many barriers to CRCS through equal access to health care and the provision of screening and preventive services for low-income veterans. When controlling for age, sex, race, and educational level, the NHIS study found that lack of insurance coverage and a usual source of care are negative factors in screening.22 On the basis of VA health-care system priorities, the covered population is enriched for those living in poverty. California state CRCS 2002 data from the American Cancer Society found that 42.8% of the target population reported a lower endoscopy within the past 5 years.21 For persons in poverty, the rate decreased to 24.6%.21 Another factor in CRCS compliance is physician recommendation.27,29 The VA screening directive raised the awareness of primary-care providers and increased the accountability of clinical and administrative decision makers. Making screening a monitored performance measure and a factor of institutional evaluation increased the incentive for facilities to screen patients.
The screening successes and stage migration seen at PAVAHCS have potential implications for improved survival in the veteran population served. It is known that 5-year survival improves significantly for early-stage CRC. Furthermore, treatment regimens are based on the stage at presentation. Morbidity associated with adjuvant therapies and significant surgery is avoided in earlier cancers and especially for curative polypectomies. A study comparing CRC survival in VA patients with national data from 1987 to 2001 showed a greater survival improvement for veterans over that period compared with national data.30 Although this survival benefit was maintained even within a specific CRC stage at presentation, higher rates of screening and awareness in the VA were proposed as possible factors in the superior VA performance.30
Stage migration has cost-savings potential for the VA health-care system. Stage migration to earlier cancers should reduce costs: costs per case increase substantially with higher-stage disease because of the need for chemotherapy and radiation. The high rate of stage 0 disease found at PAVAHCS, with 28 of the 35 stage 0 cancers treated by curative polypectomy, avoids the costs of surgery. The significant decrease in lymph node–positive stage 3 cancers avoids costs associated with adjuvant therapy. Savings may also be realized by detecting cancers on routine screening rather than through the work-up of presenting symptoms. Indeed, prediagnosis costs per patient for screen-detected CRC have been found to be less than costs for cancers detected by symptoms.31 Because the VA serves a captured health-care population, the system not only has the potential to successfully implement system-wide screening strategies, as seen for CRCS, but also has the potential to directly benefit from the cost savings associated with successful disease screening.
It is interesting to note that there was a higher percentage of stage 4 cancers at the PAVAHCS compared with the SEER and NCDB benchmarks. The average age of patients with stage 4 disease in our series was 64.7 ± 9.5 years, which is slightly younger than the average age across all stages of 67.2 years, although this difference was not statistically significant. One possible explanation is the presence at the PAVAHCS of a dedicated inpatient hospice facility and oncology treatment referral program. Patients may be referred from other VA and private facilities. Once accepted, they are staged and recorded in the cancer registry as first presentations to our facility. By definition, the patients referred to either program have advanced disease. These patients are not necessarily represented in the population of primary-care enrollees in which CRCS rates were measured. This constant source of patients with terminal disease may have contributed to the greater proportion of stage 4 disease seen at our facility and may also contribute to the lack of significant improvement in stage 4 disease over time, because any true improvement through screening would be diluted.
Several shortcomings of our study must be addressed. CRCS rates were calculated from a random chart review of only 1% of primary-care enrollees. This small sample size may not accurately reflect true screening rates. In addition, it is unknown what screening test (FOBT, sigmoidoscopy, or colonoscopy) was used and whether any of the tests ordered were diagnostic by intent instead of screening. although we looked at rates of screening for colon and rectal cancers, we were studying rectal cancers only. Therefore, findings may not be applicable to colon cancers in general, because the three screening methods used are not equally effective in the detection of proximal and distal cancers of the large bowel. A study of subsite-specific CRC incidence and stage of disease found more local disease in the descending colon (37%) and distal colorectum (41.5%), and it postulated that this was due to screening tests such as sigmoidoscopy that preferentially screen the lower bowel.32 In addition, the age at first screening and the screening status of the rectal cancer cases in this series are not known. Inherent to a VA population study is the specific nature of the veteran population. Our sample was predominantly male, and veterans tend to have lower household incomes and increased health risk behaviors such as tobacco and alcohol use. Therefore, our study results may not be applicable to the general population of men and women at risk for CRC.
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
|
FOOTNOTES |
|---|
Received for publication March 19, 2004. Accepted for publication July 12, 2004.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
