|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2002 American Society for Clinical Oncology Age, Sex, and Racial Differences in the Use of Standard Adjuvant Therapy for Colorectal CancerByFrom the Divisions of Cancer Control and Population Sciences and Cancer Treatment and Diagnosis, National Cancer Institute, Bethesda, MD; and Department of Epidemiology, University of Iowa, Iowa City, IA. Address reprint requests to Arnold L. Potosky, PhD, Applied Research Program, National Cancer Institute, 6130 Executive Blvd, EPN Room 4005, Bethesda, MD 20892; email: potosky{at}nih.gov
PURPOSE: Dissemination of efficacious adjuvant therapies for resectable colorectal cancer has not been comprehensively described. Trends, patterns, and outcomes of adjuvant therapy for colorectal cancer, focusing on age, sex, and racial/ethnic differences, are reported. MATERIALS AND METHODS: Population-based random samples of patients diagnosed with colorectal cancer diagnosed in nine geographic areas were collected annually between 1987 and 1991 and in 1995 (n = 4,706). Data were obtained from medical record reviews. Multiple logistic regression was used to assess the use of standard adjuvant chemotherapy for colon and rectal cancers. The Cox proportional hazards model was used to assess 9-year mortality. RESULTS: From 1987 until 1995, the use of adjuvant therapy increased in all age groups. There was an increase starting in 1989 for colon and in 1988 for rectal cancer. Use of standard therapy was 78% for those younger than 55 years and 24% for those older than 80 years. White patients received standard therapy more frequently than African-Americans (odds ratio, 1.75; 95% confidence interval [CI], 1.09 to 2.83). All-cause and cancer-specific mortality exceeding 9 years were lower in those who received standard therapy (all-cause risk ratio [RR], 0.73; 95% CI, 0.61 to 0.88; cancer-specific RR, 0.87; 95% CI, 0.70 to 1.09). CONCLUSION: Standard adjuvant therapies for colorectal cancer disseminated into community practices during the 1990s. However, evidence exists of differential use of therapies by older patients and by African-Americans. The use of standard therapies in the general population is associated with lower mortality. Improved dissemination of standard adjuvant therapies to all segments of the population could help reduce mortality.
COLORECTAL CANCER is the fourth most common form of cancer and the second leading cause of cancer deaths in the United States. In 2001, an estimated 135,400 new patients will be diagnosed and 56,700 people will die of the disease.1 Certain groups bear a disproportionate share of the mortality burden: people more than 75 years old, men, and African-Americans.2 Although mortality rates are declining overall, they began to decline earlier and more rapidly in whites versus African-Americans during the past two decades.1 It is likely that group disparities in colorectal cancer mortality are partly attributable to differences in risk factors; however, variations in medical practices, primarily early detection and treatment, are also potential contributing causes of these mortality differentials. In this study, we investigate patterns in the use of adjuvant therapies with proven efficacy, examining variations according to age, sex, and race/ethnicity in a population-based sample of newly diagnosed colorectal cancer patients. For the majority of colorectal cancer patients, primary therapy consists of surgical removal of the tumor. As a result of an accumulation of evidence from multiple randomized clinical trials, combined with a National Institutes of Health (NIH) Consensus Development Conference in 1990 recommending adjuvant therapies, several chemotherapy regimens have emerged in the past decade as the minimum acceptable standard of care for patients with newly diagnosed resectable colorectal cancer.3-6 On the basis of the outcomes of several more recent trials, an updated recommendation was made at the 1997 Annual Meeting of the American Society of Clinical Oncology that any one of three regimens could be considered appropriate for patients with stage III colon cancer: (1) fluorouracil (5-FU) and levamisole for 1 year, (2) 5-FU and low-dose leucovorin for 6 months, or (3) 5-FU and high-dose leucovorin for 6 months.7-9 For patients diagnosed with stage II colon cancer, more controversy exists regarding the effectiveness of chemotherapy. One retrospective analysis suggested that the relative reduction in the risk of recurrence for patients on adjuvant chemotherapy was similar in stage II and stage III disease.8 However, a meta-analysis of 1,000 stage II colon cancer patients combining data from multiple trials indicated no significant advantage in disease-free or overall survival at 5 years when patients treated with adjuvant 5-FU plus leucovorin were compared with untreated controls.10 The 1990 NIH Consensus Conference also recommended the use of the combined modality of adjuvant chemotherapy plus high-dose external-beam radiotherapy (RT) for stages II and III rectal cancer.5 Further studies have demonstrated that 5-FU alone, versus 5-FU plus other cytotoxic agents such as semustine, when combined with RT, was an acceptable standard of care.11,12 Partly as a result of concerns about the toxicity of RT on bowel function,13 the role of RT as part of the standard of care has been investigated. Results recently reported from a randomized trial of 694 patients indicated no difference in disease-specific or overall survival according to receipt of postoperative RT in patients with rectal cancer who were given adjuvant 5-FU.14 A recent comprehensive review of studies of treatment practices and outcomes of colorectal cancer found that older age and race/ethnicity, but not sex, were associated with less frequent use of postoperative adjuvant therapy.15 The studies covered by this review were based mainly on patients diagnosed earlier than 1990, before the publication of the definitive clinical trials and the 1990 NIH Consensus Conference.3-5 Other limitations of previous studies covered by the review include samples that were not population based, lack of adjustment for detailed clinical characteristics, or not verifying the provision of office-based adjuvant therapies. By use of a population-based random sample of newly diagnosed patients with verification of therapies by the treating physician, we first describe the dissemination of adjuvant therapies for stages II and III colorectal cancer between 1987 to 1995. Then we assess multiple clinical and nonclinical characteristics associated with the use of the minimally acceptable adjuvant therapy as a standard of care, focusing on age, sex, racial/ethnic, and socioeconomic differences. We also compare the mortality of patients by receipt of standard therapy to assess whether there is a benefit consistent with what has been observed in randomized clinical trials.
Between 1987 to 1991 annually, and again in 1995, all stage II and III colorectal cancer patients in the participating registries of the National Cancer Institute’s Surveillance, Epidemiology and End-Results (SEER) program were eligible for a special patterns of care study designed to assess the extent to which adjuvant therapies were offered in community medical practices. Registries covering Connecticut, Iowa, New Mexico, Atlanta, Detroit, San Francisco–Oakland, and Seattle–Puget Sound participated each year; Los Angeles participated in 1995 only because they did not join the SEER program until 1992, and Utah did not participate in 1995. Cases were sampled by cancer site and sex. Patients with a previous diagnosis of cancer (except for nonmelanoma skin cancer) or with simultaneous cancer diagnoses were ineligible for participation in this study. There was oversampling in 1995 by race/ethnicity to obtain more stable estimates; for example, 100% of Hispanic patients were sampled, and among African-American patients, approximately 50% of colon cases and 100% of rectal cases were sampled. Twenty patients were excluded because there was no report of surgical removal of the primary tumor, and two patients were excluded who were younger than 20 years old at the time of diagnosis, leaving a total final sample size of 4,706 cases spanning 1987 to 1991 and 1995. There were 473 cases sampled in 1987, 550 in 1988, 563 in 1989, 1,249 in 1990, 757 in 1991, and 1,114 in 1995. After a case was sampled, the treating physician was identified and asked to directly verify the use of chemotherapy and RT for that patient. This method augments standard SEER registry data collection procedures that rely mostly on hospital and other facility records. Doctor verification substantially improves the completeness of chemotherapy ascertainment because adjuvant chemotherapy is frequently given outside the hospital setting, and a note of any planned chemotherapy is typically not recorded in the hospital record. In 1991 and 1995, the participating registries monitored the percentage of patients whose doctors verified the receipt of outpatient therapies. For approximately 81% of sampled cases, the offer or administration of adjuvant therapy was verified from physician contacts. Medical records of sampled patients were also abstracted to collect variables not routinely collected by the SEER program, such as the presence of comorbid conditions, participation in a clinical trial, and presence of perforation or obstruction, as well as whether the margins of resection were free of cancer. Comorbidity was assessed by the Charlson comorbidity scale, a weighted index measure of comorbidity that predicts 1-year all-cause mortality and 10-year non–breast cancer mortality.16 The index comprises 19 conditions, each of which is assigned a weight according to its potential for influencing mortality. Routinely collected SEER variables included in the analysis are year of diagnosis, age at diagnosis, stage at diagnosis, marital status, histologic type and grade, and anatomic location. The SEER program assigns a stage at diagnosis by the tumor-node-metastasis system of classification.17 Tumor extent was defined as T1 (malignancy confined to mucosa or submucosal layer of the bowel), T2 (invasion to but not through the muscle of the bowel), T3 (invasion through the bowel wall or extension to adjacent connective tissue), and T4 (invasion of or through the serosa or beyond). This study is confined to tumor-node-metastasis stage II tumors, defined as T3 or T4 with no positive regional lymph nodes, and stage III tumors, defined as any T1 to T4 with positive regional lymph nodes. These stage definitions correspond approximately to stages B2 and C of the Astler-Coller modification of Dukes’ original staging system. SEER does not collect individual measures of socioeconomic status for patients. To obtain a measure of socioeconomic status, race-specific values for group-level median household income data from the United States 1990 decennial census were merged with sampled cases by using the census tract of the residence at the time of initial diagnosis. We considered any stage III colon cancer patient offered, recommended, or administered adjuvant 5-FU plus levamisole or leucovorin as receiving standard therapy. This definition conservatively accounts for provider intention to administer the appropriate standard of care and patient refusals. For stage II colon cancer, there is insufficient consensus to define a minimal standard of care regarding adjuvant chemotherapy. Nevertheless, we included these cases in selected tables for completeness of reporting. For stage II and III rectal cancer, we consider patients to have received standard therapy if they were offered, recommended, or administered 5-FU chemotherapy, with or without RT. This definition reflects the fact that 5-FU alone may also represent an acceptable standard of care, particularly in light of the most recent trial results.14 We performed a separate analysis including combined adjuvant therapy as the standard of care for rectal cancer.
The Cox proportional hazards model was used to estimate the relative risk of all-cause and cancer-specific 9-year mortality for patients diagnosed in 1990, 1991, and 1995 among those receiving standard therapy relative to those who did not receive the standard while adjusting for clinical and nonclinical characteristics. Active follow-up by the SEER registries, which collect underlying cause of death from death certificates, was complete through December 1998, at the time of this analysis. We constructed two separate models to assess all-cause and cancer-specific mortality. For the latter model, we censored deaths where cancer was not the underlying cause. Standard log-log plots were used to evaluate covariates with respect to the proportional hazard assumption. For estimates of trends and population prevalence and in the logistic regression model, sample weights were used. These weights were calculated as the inverse of the percentage of all eligible SEER cases within sampling strata defined by year of diagnosis, site (colon or rectum), sex, and race/ethnicity. All frequency estimates and regression coefficients, and associated variances, were computed by the SUDAAN statistical package (Research Triangle Institute, Research Triangle Park, NC). Wald-type F statistics that used the robust variance estimator were used to assess the statistical significance of estimated regression coefficients. All Pvalues were two-sided.
Figure 1 illustrates trends in the adjuvant chemotherapy offered or provided to colon cancer patients from 1987 to 1995 by stage and age at diagnosis. From 1988 on, the prevalence of adjuvant chemotherapy remains higher at all ages in stage III disease, where it is the standard of care, versus stage II disease, where it is not. In stage III, the dissemination of adjuvant therapy increased starting in 1989, peaked in 1990, and then remained at similar levels afterward. These trends in both stages reveal substantial differences in the prevalence of adjuvant chemotherapy according to age at diagnosis, with dissemination highest in those aged 55 or younger (reaching 90% by 1995 in stage III) and poorest in those aged 75 and older.
Figure 2 illustrates trends in different adjuvant therapies offered or given to rectal cancer patients from 1987 to 1995. The use of combined adjuvant chemotherapy with RT began to substantially increase starting in 1989, coinciding with a symmetric decline in the provision of adjuvant RT alone. The use of 5-FU chemotherapy without radiation began increasing in 1988, although dissemination has been much slower for chemotherapy alone than for combined adjuvant therapy. Prevalence of chemotherapy with or without RT is greater at all ages in patients with stage III compared with stage II disease.
Figure 3 illustrates trends in the use of adjuvant therapy for rectal cancer, defined as 5-FU–based chemotherapy with or without RT, by age group. More rapid dissemination has been observed in stage III versus stage II disease across all age groups. The increase in adjuvant therapy began in 1988 for stage III disease and in the youngest age group in stage II disease. Dissemination occurred earliest and most dramatically in the age group younger than 55 years, where a plateau was reached by 1989 in stage III disease and 1991 in stage II disease. Among rectal cancer patients older than 75 years, dissemination may be continuing after 1995.
On the basis of the trend analysis, it is apparent that most of the dissemination of standard adjuvant therapy occurred after 1989; therefore, we included patients diagnosed in 1990, 1991, and 1995 for subsequent cross-sectional analyses of patient characteristics associated with standard therapies. Table 1 lists the characteristics of the study sample used for modeling the receipt of therapy and survival outcomes. Excluded are 88 colon and 33 rectal cancer patients for whom the offer or receipt of chemotherapy was unknown. Stage II colon cancer cases (n = 841) are also excluded because there was no accepted standard of care regarding adjuvant therapy. For subsequent analyses, there are 779 stage III colon cancer cases, and 617 stage II and 749 stage III rectal cancer cases diagnosed in 1990 to 1991 or 1995. When weighted for the sampling design, these cases represent 3,965 stage III colon and 2,813 stage II or III rectal cancer patients eligible for the study.
We conducted logistic regression analyses to assess the contribution of multiple patient clinical and nonclinical characteristics toward the receipt of standard adjuvant chemotherapy; we used our definitions for standard therapy in each cancer-stage group (Table 2). In this population, age is the most important patient characteristic affecting the receipt of standard therapy, after adjustment for all other factors. The lowest use of standard therapy occurred among patients aged 75 and older (65% in 65- to 74-year-olds to 47% in 75- to 79-year-olds, and 24% in those 80 years and older). Standard therapy was received among 78% of patients younger than 55 years old. Adjustment for all other characteristics in Table 2 did not alter the age-therapy association.
After adjustment for all other variables, 58% of whites compared with 47% of African-Americans received standard care (OR, 1.75; 95% confidence interval [CI], 1.09 to 2.83). Hispanics also received standard care more frequently than did African-Americans, although the OR overlapped unity after adjustment (OR, 1.75; 95%, CI 0.88 to 3.48). The difference between white and African-Americans was smaller in patients younger than 55 years old than in patients aged 65 to 80 years. However, we examined, but did not find, a statistically significant interaction of age and race with respect to the receipt of standard therapy (data not shown). After adjustment for sociodemographic and clinical characteristics, women were more likely than men to receive standard therapy (adjusted percentages, 62% women, 53% men; OR, 1.59; 95% CI, 1.17 to 2.15). Marital status was also associated with standard therapy after adjustment for other characteristics, with patients who were unmarried (adjusted, 51%) less likely than married people to receive standard adjuvant therapy (adjusted, 61%). The effects of marital status on receipt of the standard were consistent in both men and women (data not shown). Geographic region was, overall, significantly associated with the use of standard care (P = .02), with adjusted percentages ranging from 44% in New Mexico to 65% in Iowa. However, the risk ratios indicate that no single region differed significantly from the referent group (San Francisco–Oakland). After adjustment, rectal versus colon cancer and positive nodal status were both associated with receipt of standard therapy. In the model, nodal status serves as a proxy for stage at diagnosis because by definition, stage II disease has no positive nodes. Patients with more poorly differentiated histologic grade, who were thus at higher risk for recurrent disease, were also more likely to receive standard therapy than patients with well-differentiated tumors.
The presence of comorbid conditions (Charlson comorbidity scale) was inversely significantly associated with the receipt of standard therapy. Patients with We also examined, but did not find, associations between standard therapy and several other characteristics, including insurance status (public v private coverage), the presence of bowel perforation or obstruction, tumor size, synchronous colorectal tumors, histologic type (mucinous or signet ring v carcinoma), or positive surgical margins of resection. We also examined multiple interaction terms among age, sex, race, and various tumor characteristics but detected no significant effect modification.
Table 3 lists risk ratios for all-cause and cancer-specific mortality. After adjustment for all other patient characteristics, all-cause mortality during 9 years was lower in the standard therapy group compared with those not receiving the standard (risk ratio, 0.73; 95% CI, 0.61 to 0.88). Cancer-specific mortality was also lower in the standard therapy group, but the risk ratio overlapped unity (risk ratio, 0.87; 95% CI, 0.70 to 1.09). We examined, but did not detect, any significant interactions between receipt of therapy and the other covariates. Other variables significantly associated with worse survival in this cohort included being older, being unmarried, having T4 disease, having
For stage III colorectal and stages II and III rectal cancers, which account for more than half of all new cases, or approximately 70,000 cases annually, adjuvant therapies are available that have been proven in multiple randomized trials to increase disease-free and overall survival. Improved dissemination of these therapies and future therapeutic advances to all segments of the population may be crucial for reducing mortality rates for the second largest cause of cancer deaths in the United States. The trends analysis demonstrates that the dissemination of standard adjuvant therapy for colorectal cancer rapidly increased starting in 1989 to 1990, coinciding with the publication of major findings from large, randomized trials and the publication of the recommendations of the NIH Consensus Conference.3-5 After the rapid increase in use, there was some leveling off in the use of standard therapy from 1991 to 1995, particularly for colon cancer. Our findings of differences in receipt of standard therapies according to age, sex, race, marital status, and geographic region reflect a potentially serious shortfall in the delivery of the highest possible standard of care to all segments of the population. Addressing this gap might help reduce known disparities in mortality and stage-specific relative survival rates across age, sex, and racial/ethnic groups. The mortality rates for colorectal cancer increase significantly with age.2 Also, estimates of 5-year relative survival are approximately 63% for patients diagnosed when they are younger than age 75 years, but 58% for patients diagnosed at age 75 years and older.19 One possible contributing factor to this age differential in relative survival is underuse of effective adjuvant therapies. Our findings are consistent with a recent study demonstrating declining use of 5-FU adjuvant therapy for colon cancer patients aged 65 years and older in the Medicare Program.20 The sharp decline in the use of standard adjuvant therapy starting at age 75 years, even after adjusting for tumor severity and comorbidity, raises the possibility of age bias in the management of the disease. Because our measure of standard therapy included patients whose doctors recommended they receive standard therapy and who either refused therapy or whose therapy use was unknown, this gap in care is not likely to be entirely explained by declining patient compliance with age. Receptivity to undergoing chemotherapy by the elderly has not been extensively investigated. Some doctors may not endorse aggressive therapy for elderly cancer patients because they do not expect the gains in overall survival to outweigh the potential costs in terms of patient morbidity and quality of life. Observed regional variations in the use of standard care in our results suggest the existence of differing physician practice styles or opinions related to these uncertainties about the risk-benefit tradeoffs of aggressive treatment. Much of this uncertainty may have been caused by the systematic exclusion of elderly patients from clinical trials, at least until recent years. The persistence of "ageism" in cancer care may rest on incorrect assumptions about the ability of elderly patients to withstand the toxicity of treatments. For example, a meta-analysis of phase III randomized trials demonstrates that elderly patients benefit from 5-FU adjuvant therapy without a significant increase in toxicity relative to younger patients.21 Another study found no significantly higher toxicity after adjuvant chemotherapy in colon cancer patients older than age 70 years versus younger patients.22 The inclusion of more elderly patients in clinical trials of cancer care may partly help to reinforce these findings regarding toxicity and may ultimately help to increase the numbers of patients who are receiving effective therapies. We found that women receive standard therapy more often than men, in contrast to prior studies that have found few, if any, differences in receipt of adjuvant therapy or stage-specific survival according to sex.15 Although men have higher age-adjusted mortality rates than women, the mortality differences are likely the result of many reasons, including differences in incidence, screening, and treatment. Reasons for our findings regarding sex are unclear and require further investigation. One possibility is that women more actively seek adjuvant therapy, either through questions to or suggestions from their doctors or consultations with other doctors. We also found a strong association between being married and receipt of standard therapy, an effect that remained consistent by age and sex. The role of one’s spouse in seeking more aggressive treatment and providing more support, in terms of making and keeping appointments with specialists, is an interesting research issue that deserves closer attention. Our findings of racial differences in receipt of standard adjuvant therapy are troublesome, particularly in light of African-Americans having higher mortality rates than whites (22.8 per 100,000 population v 16.8 per 100,000 population) and poorer 5-year relative survival rates for stages II and III colorectal cancer (80% in whites v 71% for African-Americans in stage II, and 57% whites and 53% in African-Americans for stage III).19 Among Hispanics, 5-year relative survival rates are more similar to whites than to African-Americans. The stage-specific relative survival differences may be partly attributable to differences in the receipt of standard adjuvant therapy observed in this study. We found the use of standard therapy statistically significantly higher in whites compared with African-Americans, after adjustment for other patient characteristics. Our findings are consistent with an analysis of colon cancer treatment in 1996 in the Carolinas that found significantly more adjuvant therapy in whites compared with African-Americans.23 An analysis comparing outcomes between African-Americans and whites participating in colon cancer adjuvant therapy trials found similar efficacy by race, suggesting that adjuvant therapy could help to improve prognosis for African-Americans in the general community.24 The observed differences between African-Americans and whites in the receipt of standard therapy narrowed between 1990 to 1991 and 1995, but this trend was not statistically significant after adjustment for all other patient clinical and nonclinical factors. The underlying reasons for differences in receipt of standard therapy by race may still persist in more recent years, or we may have insufficient annual sample sizes to detect different trends by race in the receipt of standard therapy. Recent data collection efforts by SEER to assess the receipt of standard therapy by colorectal cancer patients diagnosed in 2000 will help to determine whether these racial/ethnic differences in standard therapy have continued. The lower use of standard adjuvant therapy by African-Americans may be the result of many factors, such as the use of different providers with variable practice styles or racial bias in provision of treatments. However, the measures available limit our ability to identify these or to specify other reasons for racial disparities in cancer care. For example, our use of public versus private insurance coverage may be too general to capture the gradations of health care coverage that might differ by race group and that are associated with differential delivery of therapies for colorectal cancer.25 We did not have available individual measures of socioeconomic status, such as income or occupation, and census-tract measures should not be considered as proxies for individual measures.26 However, we included group-level socioeconomic status variables as a contextual measure because such measures have been associated with other health outcomes independent of individual measures.27,28 Finally, we were unable to account for potentially important differences in baseline health status at diagnosis, or health status after surgical resection of the primary tumor, that might confound age and race associations with receipt of postoperative adjuvant therapy. The inclusion of comorbidity may not be sufficient to account for differences in health status by age or race that may influence provision of adjuvant therapies. We found that both overall and cancer-specific mortality were lower in patients who received standard adjuvant therapy, after adjustments for multiple clinical and nonclinical factors, which suggests a mortality benefit from standard therapies among patients treated in community settings. Our results indicate an approximate 27% overall 9-year mortality reduction (P < .01). However, the 13% reduction in cancer-specific mortality was not statistically significant at an alpha level of 0.05. The more pronounced effect of standard adjuvant therapy on overall mortality may be due primarily to unmeasured differences between the two groups with respect to underlying health status. Any comparisons of mortality in an observational study must be cautiously interpreted, given the potential for selection bias; for example, healthier patients at baseline with fewer comorbidities tend to be those who receive more intensive cancer care. To assess and reduce the effects of heterogeneity of the two comparison groups, we conducted a propensity score analysis of mortality following the method of Rosenbaum and Rubin.29 The propensity score was defined as the probability of a patient receiving standard adjuvant therapy adjusting for all measured baseline clinical and sociodemographic characteristics and was derived by the same model as used in Table 2. The propensity scores were then used to assess whether there was sufficient overlap of covariates between the two treatment groups to justify comparing their mortality outcomes. After determining there was sufficient overlap, the association of mortality with adjuvant therapy was then evaluated, adjusting for the propensity score. This method has been applied to compare outcomes among treatment groups in various other observational settings, including heart disease and obstetric care.30,31 By use of this method, similar mortality risk ratios for standard therapy were obtained as those reported in Table 3. Although no analytic method can completely remove selection bias, the mortality results are generally consistent with findings from randomized clinical trials that tend to enroll more highly selected patients than those included in our population-based sample. This provides more evidence in support of wider dissemination of these therapies in the population. The nine geographic areas included in our study may not be representative of the entire United States with respect to medical practices and the sophistication of delivery of standard cancer therapy. We also did not account for potentially important provider characteristics, such as experience and volume, referral patterns, access to medical and radiation oncologists, and structural and financial arrangements of health care delivery systems. Our definition of standard therapy for rectal cancer may be too broad; during the early 1990s, chemotherapy combined with RT may have been considered the standard as recommended by the NIH Consensus Conference. A total of 57% of all eligible patients in Table 2 received the standard when defined as chemotherapy with or without RT for rectal cancer, but this was reduced to 50% when more narrowly defined as combined adjuvant chemoradiotherapy. To assess the effect of changing the standard, we reran the logistic regression model that produced results in Table 2 and used the more restrictive definition of standard adjuvant therapy (chemoradiotherapy). The main associations reported in Table 2 persisted, but some ORs for receipt of standard care changed. For example, the OR for women changed from 1.59 to 1.35, and for whites versus African-Americans, the OR changed from 1.75 to 1.58. The inability to directly verify the receipt of outpatient therapy with treating physicians in 19% of the sample might potentially confound our results. To assess the possibility of a differential tendency to verify therapy according to the independent variables in our model, we added a covariate term to the logistic regression model in Table 2 representing verification of therapy. Although this variable was strongly independently associated with receipt of standard therapy, the associations of the other model variables in the model with receipt of standard therapy did not substantially change. In conclusion, our results demonstrate a relatively rapid dissemination of standard colorectal cancer adjuvant therapies into community practices coinciding with the publication of results from breakthrough clinical trials and the NIH Consensus Development Statement, with a subsequent leveling in the early 1990s. However, there is evidence of the differential provision of these therapies, especially to older patients and to African-Americans. Given our findings of improved mortality among those receiving the standard of care in the community setting consistent with results from major clinical trials, improved dissemination of standard therapies for colorectal cancer to all segments of the population should be a goal of researchers, policy makers, and health care providers.
1. Greenlee RT, Hill-Harmon MB, Murray T, et al: Cancer statistics, 2001. CA Cancer J Clin 51: 15-36, 2001 2. Ries LAG, Wingo PA, Miller DS, et al: The annual report to the nation on the status of cancer, 1973-1997, with a special section on colorectal cancer. Cancer 88: 2398-2424, 2000[CrossRef][Medline] 3. Moertel CG, Fleming TR, Macdonald JS, et al: Levamisole and fluorouracil for adjuvant therapy of resected colon carcinoma. N Engl J Med 322: 352-358, 1990[Abstract] 4. Krook JE, Moertel CG, Gunderson LL, et al: Effective surgical adjuvant therapy for high-risk rectal carcinoma. N Engl J Med 324: 709-715, 1991[Abstract]
5.
NIH Consensus Conference: Adjuvant therapy for patients with colon and rectal cancer. JAMA 264: 1444-1450, 1990 6. Macdonald JS: Adjuvant therapy for colon cancer. CA Cancer J Clin 47: 243-256, 1997[Abstract]
7.
Wolmark N, Rockette H, Mamounas E, et al: Clinical trial to assess the relative efficacy of fluorouracil and leucovorin, fluorouracil and levamisole, and fluorouracil, leucovorin, and levamisole in patients with Dukes’ B and C carcinoma of the colon: Results from National Surgical Adjuvant Breast and Bowel Project C-04. J Clin Oncol 17: 3553-3559, 1999
8.
Mamounas E, Wieand S, Wolmark N, et al: Comparative efficacy of adjuvant chemotherapy in patients with Dukes’ B versus Dukes’ C colon cancer: Results from four National Surgical Adjuvant Breast and Bowel Project adjuvant studies (C-01, C-02, C- 03, and C-04). J Clin Oncol 17: 1349-1355, 1999
9.
O’Connell MJ, Laurie JA, Kahn M, et al: Prospectively randomized trial of postoperative adjuvant chemotherapy in patients with high-risk colon cancer. J Clin Oncol 16: 295-300, 1998
10.
International Multicentre Pooled Analysis of B2 Colon Cancer Trials (IMPACT B2) Investigators: Efficacy of adjuvant fluorouracil and folinic acid in B2 colon cancer. J Clin Oncol 17: 1356-1363, 1999 11. Kollmorgen CF, Meagher AP, Wolff BG, et al: The long-term effect of adjuvant postoperative chemoradiotherapy for rectal carcinoma on bowel function. Ann Surg 220: 676-682, 1994[Medline]
12.
O’Connell MJ, Martenson JA, Wieand HS, et al: Improving adjuvant therapy for rectal cancer by combining protracted-infusion fluorouracil with radiation therapy after curative surgery. N Engl J Med 331: 502-507, 1994
13.
Gastrointestinal Tumor Study Group: Radiation therapy and fluorouracil with or without semustine for the treatment of patients with surgical adjuvant adenocarcinoma of the rectum. J Clin Oncol 10: 549-557, 1992
14.
Wolmark N, Wieand HS, Hyams DM, et al: Randomized trial of postoperative adjuvant chemotherapy with or without radiotherapy for carcinoma of the rectum: National Surgical Adjuvant Breast and Bowel Project Protocol R-02. J Natl Cancer Inst 92: 388-396, 2000
15.
Hodgson DC, Fuchs CS, Ayanian JZ: Impact of patient and provider characteristics on the treatment and outcomes of colorectal cancer. J Natl Cancer Inst 93: 501-515, 2001 16. Charlson ME, Pompei P, Alex KL, et al: A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis 40: 373-383, 1987[CrossRef][Medline] 17. American Joint Committee on Cancer: Manual for Staging of Cancer (ed 3). Philadelphia, PA, Lippincott, 1988 18. Graubard BI, Korn EL: Predictive margins with survey data. Biometrics 55: 652-659, 1999[CrossRef][Medline] 19. Ries LAG, Eisner MP, Kosary CL, et al: SEER cancer statistics review 1997. SEER-Stat version 4.0. Accessed April 1, 2001. Http://seer.cancer.gov/
20.
Schrag D, Cramer LD, Bach PB, et al: Age and adjuvant chemotherapy use after surgery for stage III colon cancer. J Natl Cancer Inst 93: 850-857, 2001 21. Sargent DR, Goldberg J, MacDonald J, et al: Adjuvant chemotherapy for colon cancer is beneficial without significantly increased toxicity in elderly patients: Results from a 3,351 patient meta-analysis. Proc Am Soc Clin Oncol 19: 241a, 2000 (abstr 933) 22. Aschele C, Guglielmi A, Tixi LM, et al: Adjuvant treatment of colorectal cancer in the elderly. Cancer Control 2: 36-38, 1995 (suppl 1)[Medline] 23. Tropman SE, Hatzell T, Paskett E, et al: Colon cancer treatment in rural North and South Carolina. Cancer Detect Prev 23: 428-434, 1999[CrossRef][Medline]
24.
Dignam JJ, Colangelo L, Tian W, et al: Outcomes among African-Americans and Caucasians in colon cancer adjuvant therapy trials: Findings from the National Surgical Adjuvant Breast and Bowel Project. J Natl Cancer Inst 91: 1933-1940, 1999
25.
Roetzheim RG, Pal N, Gonzalez EC, et al: Effects of health insurance and race on colorectal cancer treatments and outcomes. Am J Public Health 90: 1746-1754, 2000
26.
Geronimus AT, Bound J: Use of census-based aggregate variables to proxy for socioeconomic group: Evidence from national samples. Am J Epidemiol 148: 475-486, 1998
27.
Krieger N: Overcoming the absence of socioeconomic data in medical records: Validation and application of a census-based methodology. Am J Public Health 82: 703-710, 1992 28. Greenwald HP, Polissar NL, Borgatta EF, et al: Detecting survival effects of socioeconomic status: problems in the use of aggregate measures. J Clin Epidemiol 47: 903-909, 1994[CrossRef][Medline] 29. Rosenbaum P, Rubin DB: Reducing bias in observational studies using subclassification on the propensity score. JASA 79: 516-524, 1984
30.
Connors AF Jr, Speroff T, Dawson NV, et al: The effectiveness of right heart catheterization in the initial care of critically ill patients. JAMA 276: 889-897, 1996 31. Lieberman E, Lang JM, Cohen A, et al: Association of epidural analgesia with cesarean delivery in nulliparous women. Obstet Gynecol 88: 93-100, 1996[CrossRef][Medline] Submitted July 20, 2001; accepted November 7, 2001.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2002 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
2 tests were used to evaluate bivariate relationships between patient characteristics and receipt of standard care, within each cancer site–stage combination. A two-tailed test was used. A logistic regression model was constructed to assess the association between each independent variable and the receipt of standard care. We report the adjusted percentages of patients receiving standard therapy in addition to the odds ratios (ORs) because the latter may overstate the risk ratio when the outcome is not a rare event. To obtain the adjusted percentages, the logistic regression models were used to generate estimates of the probability for each individual (or predicted values from the models) receiving standard therapy. The percentages in each group were then directly standardized to the distribution of the covariates among the entire weighted sample.
two other medical conditions were less likely to receive standard therapy than patients without any comorbidities after adjustment for all other factors, including age. There was no significant interaction effect of age and comorbidity (data not shown). 
