Search tips
Search criteria 


Logo of jnciLink to Publisher's site
J Natl Cancer Inst. 2012 February 8; 104(3): 211–227.
Published online 2012 January 20. doi:  10.1093/jnci/djr524
PMCID: PMC3274510

Comparative Effectiveness of Oxaliplatin vs Non–Oxaliplatin-containing Adjuvant Chemotherapy for Stage III Colon Cancer



The addition of oxaliplatin to adjuvant 5-fluorouracil (5-FU) improves survival of patients with stage III colon cancer in randomized clinical trials (RCTs). However, RCT participants are younger, healthier, and less racially diverse than the general cancer population. Thus, the benefit of oxaliplatin outside RCTs is uncertain.

Subjects and Methods

Patients younger than 75 years with stage III colon cancer who received chemotherapy within 120 days of surgical resection were identified from five observational data sources—the Surveillance, Epidemiology, and End Results registry linked to Medicare claims (SEER–Medicare), the New York State Cancer Registry (NYSCR) linked to Medicaid and Medicare claims, the National Comprehensive Cancer Network (NCCN) Outcomes Database, and the Cancer Care Outcomes Research & Surveillance Consortium (CanCORS). Overall survival (OS) was compared among patients treated with oxaliplatin vs non–oxaliplatin-containing adjuvant chemotherapy. Overall survival for 4060 patients diagnosed during 2004–2009 was compared with pooled data from five RCTs (the Adjuvant Colon Cancer ENdpoinTs [ACCENT] group, n = 8292). Datasets were juxtaposed but not combined using Kaplan–Meier curves. Covariate and propensity score adjusted proportional hazards models were used to calculate adjusted survival hazard ratios (HR). Stratified analyses examined effect modifiers. All statistical tests were two-sided.


The survival advantage associated with the addition of oxaliplatin to adjuvant 5-FU was evident across diverse practice settings (3-year OS: RCTs, 86% [n = 1273]; SEER–Medicare, 80% [n = 1152]; CanCORS, 88% [n = 129]; NYSCR–Medicaid, 82% [n = 54]; NYSCR–Medicare, 79% [n = 180]; and NCCN, 86% [n = 438]). A statistically significant improvement in 3-year overall survival was seen in the largest cohort, SEER–Medicare, and in the NYSCR–Medicare cohort (non–oxaliplatin-containing vs oxaliplatin-containing adjuvant therapy, adjusted HR of death: pooled RCTs: HR = 0.80, 95% CI = 0.70 to 0.92, P = .002; SEER–Medicare: HR = 0.70, 95% CI = 0.60 to 0.82, P < .001; NYSCR–Medicare patients aged ≥65 years: HR = 0.58, 95% CI = 0.38 to 0.90, P = .02). The association between oxaliplatin treatment and better survival was maintained in older and minority group patients, as well as those with higher comorbidity.


The addition of oxaliplatin to 5-FU appears to be associated with better survival among patients receiving adjuvant colon cancer treatment in the community.


Prior knowledge

Randomized controlled trials have demonstrated that oxaliplatin-containing adjuvant therapy improves the survival of patients with stage III colon cancer. However, it is not known whether this therapy improves survival to the same extent in cancer patients in the general population.

Study design

The overall survival of 4060 patients treated with oxaliplatin-containing vs non–oxaliplatin-containing adjuvant chemotherapy was assessed from five observational data sources and compared with pooled data from five randomized controlled trials.


The use of oxaliplatin-containing adjuvant therapy was associated with a consistent pattern of improved survival across the diverse practice settings represented by the five cohorts of stage III colon cancer patients, including older and minority group patients and those with higher comorbidity.


The addition of oxaliplatin to adjuvant therapies for stage III colon cancer shown to confer a survival advantage by randomized controlled trials has proven to be equally beneficial in the more diverse therapeutic settings in the general population.


The small sample sizes of racial and ethnic minorities limited the precision of the hazard ratio estimates. Different follow-up times in the different cohorts precluded a unified interpretation across all cohorts. The potential for confounding based on patient selection as a substantial cause of improved survival in oxaliplatin-treated patients could not be eliminated.

From the Editors

Colon cancer is a leading cause of morbidity and mortality worldwide. An anticipated 101 340 Americans will be diagnosed with colon cancer in 2011, about one-third of whom will have stage III, or node-positive, disease (1,2). Surgery is the mainstay of curative therapy for stage III colon cancer. Surgery alone, however, results in an unacceptably low 5-year disease-free survival of 15%–50% depending on substage (3). During 1990–2004, post-surgical adjuvant chemotherapy with leucovorin-modulated fluorouracil (hereafter 5-FU) was the standard of care for stage III colon cancer based on a 26% relative reduction in mortality compared with surgery alone (3,4). In 2004, the Multicenter International Study of Oxaliplatin/5-Fluorouracil/Leucovorin in the Adjuvant Treatment of Colon Cancer (MOSAIC) trial reported that the addition of oxaliplatin to 5-FU provided 23% further improvement in relative disease-free survival and 4% absolute improvement in 6-year overall survival in stage III colon cancer (5,6). The National Surgical Adjuvant Breast and Bowel Project (NSABP) C-07 trial confirmed this finding (7), making adjuvant therapy with 5-FU and oxaliplatin the new standard for patients with resected stage III colon cancer.

The addition of oxaliplatin to 5-FU unequivocally improves outcomes of patients enrolled in RCTs; however, fewer than 2% of patients with incident cancer enroll in National Cancer Institute trials (8). The patient populations in clinical trials, including MOSAIC and C-07, are substantially younger, healthier, and less racially and/or ethnically diverse than the general population of cancer patients. In addition, dosing and follow-up are defined in trials but are at the discretion of the treating physician in routine practice. Because oxaliplatin increases toxicity, specifically nausea, vomiting, neutropenia, and peripheral neuropathy (5), variations in dosing might be expected to alter tolerance and drug delivery and thus attenuate the benefit of adjuvant oxaliplatin in community settings. Given these potential differences, it is critical to understand the community-based effectiveness of adjuvant chemotherapy. We sought to assess the benefit of adding oxaliplatin to 5-FU for stage III colon cancer in patients treated in the community (effectiveness) and to compare that with oxaliplatin benefit in phase III RCTs (efficacy). To that end, we compared overall survival of patients treated with oxaliplatin-containing adjuvant chemotherapy with overall survival of patients treated with non–oxaliplatin-containing regimens in effectiveness samples drawn from five observational cohorts and an efficacy sample drawn from pooled individual patient data from five phase III adjuvant chemotherapy trials.


Data Sources

Effectiveness samples came from the following sources: Surveillance, Epidemiology, and End Results registry linked to Medicare claims (SEER–Medicare); the New York State Cancer Registry (NYSCR) linked to Medicaid and Medicare claims; the National Comprehensive Cancer Network (NCCN) Outcomes Database; and the Cancer Care Outcomes Research & Surveillance Consortium (CanCORS). The efficacy sample was drawn from the Adjuvant Colon Cancer ENdpoinTs (ACCENT) group. This study was approved by the Institutional Review Boards at the University of North Carolina (IRB 09-0764); the Brigham and Women’s Hospital/Dana-Farber Cancer Institute (IRB 08-338); and the Mayo Clinic (IRB 531-04, 120-2005).

Effectiveness Samples.

The National Cancer Institute (NCI) SEER program of cancer registries collects data on incident cancer cases from registries covering 26% of the US population. Linkage of case patients to their corresponding Medicare claims allows investigation of treatment and outcomes (9,10). A novel linkage between the NYSCR and Medicaid and Medicare claims captures information on incident cancer case patients diagnosed in New York State. The NCCN Outcomes Database comprises prospectively abstracted information for incident colorectal cancers treated at eight NCI-designated comprehensive cancer centers (11,12). CanCORS is a population-based cohort study of patients with newly diagnosed colorectal cancer in four geographical regions, five large health maintenance organizations, and 15 Veterans Administration hospitals. Minorities were oversampled, and information was gathered through medical records abstraction and patient and provider surveys (13,14). There is potential overlap between effectiveness cohorts in select geographical regions; however, confidentiality restrictions precluded our ability to determine the extent of this potential overlap. Because the Food and Drug Administration approved oxaliplatin for stage III colon cancer in 2004, case patients diagnosed in 2004 or later were included.

Efficacy Sample.

ACCENT contains individual patient data from 21 phase III adjuvant colon cancer chemotherapy trials during 1977–2008. For a contemporary comparison, only trials enrolling patients in 1999 and later for which follow-up data were complete were included. Five RCTs with a 5-FU control arm met these criteria. The experimental arms consisted of capecitabine (XACT), 5-FU/irinotecan combinations (PETACC-3, C89803), and oxaliplatin/5-FU combinations (MOSAIC, C-07) (Table 1) (5,7,1517).

Table 1
Characteristics of clinical trials included as the efficacy sample from the adjuvant colon cancer ENdpoinTs database*

Sample Eligibility

RCTs have included too few patients older than 75 years to make robust conclusions about oxaliplatin's efficacy in older patients. Therefore, this analysis is restricted to patients younger than 75 years for whom RCTs compellingly demonstrate oxaliplatin benefit. All patients in all samples had histologically confirmed stage III adenocarcinoma of the colon (not rectum) resected within 90 days of diagnosis and received adjuvant chemotherapy within 120 days of resection (Figure 1). To assure complete claims information, patients in SEER–Medicare and NYSCR–Medicare were excluded if enrolled in a health maintenance organization or not continuously enrolled in both Medicare Parts A and B in the 6 months following diagnosis. To facilitate interpretation of the NYSCR with regard to dual Medicare/Medicaid eligibility, the cohort was divided into Medicaid patients younger than 65 years (32% of whom were also enrolled in Medicare) and Medicare patients who were at least 65 years of age (28% of whom were also enrolled in Medicaid).

Figure 1
Effectiveness cohort assembly. CanCORS = Cancer Care Outcomes Research & Surveillance Consortium; NCCN = National Comprehensive Cancer Network Outcomes Database; NYSCR = New York State Cancer Registry; SEER = Surveillance, Epidemiology, and End ...

Treatment Ascertainment

Two treatment groups were compared—those receiving oxaliplatin-containing chemotherapy and those receiving non–oxaliplatin-containing chemotherapy. In the effectiveness sample, the oxaliplatin group included patients with any claim/record of oxaliplatin within 30 days of their first chemotherapy dose. The non-oxaliplatin group comprised all other patients, including those who received capecitabine (1620). In the efficacy sample the oxaliplatin group comprised the oxaliplatin/5-FU arms of C-07 (n = 862) and MOSAIC (n = 672). The 5-FU arms of C-07 and MOSAIC and all arms of the other studies comprised the non-oxaliplatin group.

Statistical Analysis


Data on age (measured continuously in all cohorts except CanCORS where categories were ages 21–49, 50–64, 65–69, and 70–74 years), sex, race (white, black, Asian, Latino, other) tumor substage (IIIA, IIIB, IIIC, III not otherwise specified), tumor grade (well or moderately differentiated, poorly or undifferentiated, differentiation unknown), and year of diagnosis (2004–2009) were available in all effectiveness samples. Income (quantiles) based on zip code or census tract residence was available for SEER–Medicare, NCCN, and NYSCR; CanCORS contains individual estimates (>$60 000; $40 000–$60 000; $20 000–$40 000; <$20 000). Comorbidity was measured using the Charlson Comorbidity Index (CCI) in NCCN, its Deyo modification in NYSCR, and its Deyo-Klabunde modification in SEER–Medicare (2123). Comorbidity in CanCORS was measured using the Adult Comorbidity Evaluation (ACE)-27 index (24,25). Income and comorbidity, important covariates estimated differently in each sample, were retained in statistical analyses.

For ACCENT, age, sex, substage, and year of trial enrollment were available on all patients. Individual patient performance status and race were not available on all patients; missing data were imputed based on available clinical variables as previously reported (n = 4246) (3,26,27). Because 97% of patients enrolled in XACT and the overwhelming majority in PETACC-3 were white, all XACT and PETACC-3 patients were assumed to be white.

Overall Survival.

Overall survival in the effectiveness cohorts was measured from 30 days after surgery. This anchor date was selected because it could be reliably ascertained in the effectiveness datasets and is a comparable start point to clinical trial randomization date. Overall survival in the efficacy cohort was measured from time of random assignment to death from any cause. Because of short follow-up in most cohorts, 3-year rather than 5-year overall survival is reported.

In SEER–Medicare, NYSCR, and NCCN, vital status was ascertained through the National Death Index. For CanCORS, each participating site updated vital status with censoring at the time of each update.

Given their heterogeneity, effectiveness data were not combined but juxtaposed after applying consistent inclusion criteria and covariate specifications. Within each effectiveness cohort, univariate and multivariable logistic regression were used to assess the associations among covariates and oxaliplatin receipt. Next, overall survival of treatment groups was compared descriptively with Kaplan–Meier survival estimates. Cox proportional hazards models were used to calculate unadjusted and adjusted hazard ratios of survival. The assumption of proportionality was tested using the ASSESS option of PHREG in SAS (SAS v9.2; SAS Institute, Cary, NC). Survival in each treatment group was compared within subgroups to examine whether key covariates modified the effect of oxaliplatin. Sample size constraints precluded formal interaction testing. Adjusted proportional hazards models were also performed for the efficacy cohorts. Analyses were conducted with SAS version 9.2 (SAS Institute).

A propensity score analysis compared survival of treatment groups within a common range of risk factors for outcome in effectiveness cohorts. Using the same covariates as the proportional hazards model, we estimated the likelihood of oxaliplatin receipt (the propensity score) for each patient (28). Patients with the highest and lowest propensity for oxaliplatin without an overlapping patient of similar propensity in the other treatment group were omitted to create cohorts with overlapping propensity score distributions. Overall survival of treatment groups was compared within propensity score quintiles. Because there was no clear trend of differential treatment effect by quintile, a propensity score–adjusted proportional hazards model was used to calculate the survival hazard ratio. Disease-free survival was not evaluated because it could not be measured in all observational cohorts. All statistical tests were two-sided.


Sample Characteristics

A total of 4060 patients were in the effectiveness sample and 8292 patients in the efficacy sample. As expected from the differences in sample acquisition, there were differences in the distribution of age and race among cohorts (Table 2). Similar proportions of patients across practice settings in the effectiveness sample had serious comorbidity as defined by CCI (≥2) or ACE-27 (severe) comorbid disease (Table 2), categories shown to have comparable hazards of death (29).

Table 2
Characteristics of stage III colon cancer patients treated with adjuvant chemotherapy, by cohort*

Oxaliplatin Receipt in the Community

The proportion of patients treated with oxaliplatin varied considerably: 59% in SEER–Medicare; 54% in CanCORS; 23% in NYSCR–Medicaid; 51% in NYSCR–Medicare; and 94% in NCCN. Except in NCCN where the overwhelming majority of patients received oxaliplatin, year of diagnosis was strongly associated with oxaliplatin use, which increased substantially between 2004 and 2007 (Table 3).

Table 3
Likelihood of oxaliplatin receipt in stage III colon cancer patients treated with adjuvant chemotherapy*

The use of oxaliplatin decreased with advancing age. Among patients aged 70–74 years, 53% in SEER–Medicare, 37% in CanCORS, and 48% in NYSCR–Medicare received oxaliplatin; these rates were lower than for patients aged 65–69 years (eg, SEER–Medicare adjusted OR = 0.67, 95% CI = 0.56 to 0.81). Although 82% of 70- to 74-year-old patients treated at NCCN centers received oxaliplatin, this was still lower than the rate for 65- to 69-year-old patients (96%). Oxaliplatin use was similarly low in patients with extensive comorbidity.

The association between race and oxaliplatin receipt was not uniform. In CanCORS and NYSCR–Medicaid, there was minimal difference in likelihood of oxaliplatin use between blacks and whites, whereas in NCCN, SEER–Medicare, and NYSCR–Medicare, black patients received less oxaliplatin than whites, a difference that was substantial in both Medicare populations (eg, NYSCR–Medicare 36% vs 54%, OR = 0.44, 95% CI = 0.23 to 0.83).

Effect of Oxaliplatin on Survival

Three-year survival was superior for patients treated with oxaliplatin-containing chemotherapy across all cohorts (Figure 2, A–G and Table 4). In the efficacy sample, 3-year overall survival of oxaliplatin-treated patients was 86% (1273 patients), a 4% absolute and a 20% relative improvement in survival compared with non–oxaliplatin-treated patients (adjusted HR of death = 0.80, 95% CI = 0.70 to 0.92, P = .002; Tables 4 and and5).5). Three-year overall survival of oxaliplatin-treated patients was remarkably similar in the effectiveness cohorts: 80% (1152) SEER–Medicare, 88% (129) CanCORS, 82% (54) NYSCR–Medicaid, 79% (180) NYSCR–Medicare, and 86% (438) NCCN.

Table 4
Effect of oxaliplatin treatment on overall survival of stage III colon cancer patients treated with chemotherapy*
Table 5
Adjusted overall survival of stage III colon cancer patients treated with chemotherapy*
Figure 2Figure 2
Comparison of Kaplan–Meier survival in months by treatment in stage III colon cancer patients treated with adjuvant chemotherapy by cohort. Error bars indicate 95% confidence intervals at 10-month intervals. Cox proportional hazards models were ...

Adjusting for differences in patient and tumor characteristics between treatment arms, the survival HR point estimates of all effectiveness cohorts showed at least as much reduction in the rate of death from oxaliplatin treatment as the efficacy cohort (Table 4), although the 95% confidence interval crossed 1.0 in all but SEER–Medicare (HR of death = 0.70, 95% CI = 0.60 to 0.82, P < .001) and NYSCR–Medicare (HR of death = 0.58, 95% CI = 0.38 to 0.90, P = .02). The propensity score–adjusted hazard ratios confirmed the survival benefit from oxaliplatin seen in the covariate-adjusted models (Table 4).

The benefit of oxaliplatin was largely maintained across all clinically relevant subgroups examined, although small sample size limited the precision of hazard ratio estimates. Oxaliplatin-treated patients aged 70–74 years had improved survival in SEER–Medicare (HR of death = 0.66, 95% CI = 0.52 to 0.84) and NYSCR–Medicare (HR of death = 0.62, 95% CI = 0.36 to 1.07), but not in CanCORS (Table 4). Oxaliplatin benefit was also seen in patients with CCI of at least 2 in SEER–Medicare. In three effectiveness cohorts, black patients derived similar or greater benefit from oxaliplatin than whites, although analysis of all racial/ethnic minorities was limited by small sample size.


In this study, we found that the use of oxaliplatin-containing adjuvant therapy was associated with a consistent pattern of improved survival across the diverse practice settings represented by the effectiveness cohorts of stage III colon cancer patients, with a statistically significant improvement in the largest cohort, SEER–Medicare, and the NYSCR–Medicare cohort. The survival advantage was maintained in older, sicker, and minority group patients.

In 1990, in the wake of multiple positive RCTs, adjuvant 5-FU became the standard of care for patients with resected stage III colon cancer (4). In 2004, the addition of oxaliplatin was shown to further improve disease-free survival and subsequently overall survival (57). The oxaliplatin/5-FU combination, however, has a greater risk of severe cytopenias, nausea and vomiting, diarrhea, and peripheral neuropathy than 5-FU alone (5,7). In light of these toxic effects, which confer greater risk to oxaliplatin-treated patients, should the effectiveness of oxaliplatin be attenuated in the community, routine use of adjuvant oxaliplatin for very small reductions in mortality might not be warranted. To investigate the community effectiveness of oxaliplatin in stage III colon cancer, we assembled five observational datasets from diverse treatment settings to compare oxaliplatin's effectiveness to the efficacy demonstrated in randomized phase III trials. To mirror the trial cohorts, we restricted our analyses to patients diagnosed before age 75 years.

The use of oxaliplatin-containing chemotherapy did improve survival in these effectiveness cohorts notable for their diversity, including patients cared for across the spectrum of treatment venues in the United States: at specialty cancer centers, academic oncology groups, community oncology practices, and Veterans Administration hospitals. Despite including patients from a wide array of treatment settings, 3-year survival was very similar in oxaliplatin-treated patients in all effectiveness cohorts and the efficacy cohort trial patients. After adjusting for key factors related to treatment selection and survival, the relative reduction in death rates from the addition of oxaliplatin in all effectiveness cohorts was comparable with the 20% relative reduction seen in the efficacy data and the MOSAIC trial and the 13% relative reduction in death from capecitabine/oxaliplatin compared with 5-FU recently reported by the XELOXA trial (6,30). The relative reductions in mortality were more profound in most effectiveness cohorts than the efficacy cohort, likely related to unmeasured confounding such as frailty that led patients treated in the community to forego oxaliplatin and thereby made the non-oxaliplatin group a more frail population.

Oxaliplatin use was associated with statistically significantly improved survival in the oldest age group investigated, 70–74 years, in the SEER–Medicare cohort, with a trend toward improved survival in NYSCR–Medicare. Older CanCORS patients did not benefit from oxaliplatin, although there was no suggestion of harm. Although adjuvant 5-FU clearly improves survival among older patients, the role of combination chemotherapy has been an area of concern (14,3135). In the case of oxaliplatin/5-FU combinations, two publications (36,37) have suggested that older patients benefit from the addition of oxaliplatin, although a recent pooled analysis suggested oxaliplatin use might be harmful in patients older than 70 years (HR of death = 1.18, 95% CI = 0.90 to 1.57) (38). Although our findings do not support the notion of harm, only about half of this age group received oxaliplatin; thus, it is uncertain from this analysis whether expanding the use of oxaliplatin to the entire group of chemotherapy-treated patients aged 65–74 years would be safe or effective.

We also examined the comparative treatment effect within clinically relevant patient subgroups. Many of these subgroups were too small to yield precise estimates of oxaliplatin's effect; however, the benefit of oxaliplatin appeared to extend to patients with at least two comorbid conditions as defined by CCI in SEER–Medicare and by “moderate” comorbidity as defined by the ACE-27. The small group of oxaliplatin-treated patients with “severe” comorbidity in CanCORS had inferior survival; however, because we limited this investigation to patients who survived 30 days from resection and received their first dose of chemotherapy within 120 days, most patients with major comorbidity were excluded from the analysis.

Subgroup analysis also found that oxaliplatin is associated with a similar reduction in death rates in blacks and whites in most cohorts, although variation among cohorts merits a more extensive examination. The small sample size of all racial and ethnic minorities even in this large multi-cohort investigation suggests that efforts specifically designed to study the outcomes of minority patients with colon cancer are warranted.

This study also had some other limitations besides small sample size of racial and ethnic minorities. A multicenter, multi-cohort comparative effectiveness project presents a number of challenges in the examination and interpretation of observational data from multiple sources. For example, different follow-up times among cohorts challenged a unified interpretation. In CanCORS, the effectiveness cohort with the longest follow-up, survival curves converge around 45 months after which there appears to be no survival benefit from oxaliplatin (not shown); this was not the case in any other dataset. This convergence at a follow-up time point beyond that available in the other cohorts raises the possibility that the effect of oxaliplatin over time may not be as robust in the community as it was in the MOSAIC trial. Unfortunately, the short follow-up for most cohorts currently prevents investigation into whether oxaliplatin effect attenuates with time.

Our analysis was subject to the limitations of each data source. We limited our analyses to overall survival because it was reliably and consistently collected for each cohort; however, disease-free survival is a more direct indicator of colon cancer recurrence and thus failure of adjuvant treatment. Quality of life is another crucial metric that likely varies by the type of chemotherapy given. Neither disease-free survival nor quality of life could be reliably determined across the effectiveness cohorts.

Most importantly, notwithstanding our efforts to assemble comparable cohorts and our multivariable models, we could not eliminate the potential for confounding based on patient selection as a substantial cause of improved survival in oxaliplatin-treated patients. We used a traditional proportional hazards model to account for known relevant confounders rather than propensity score techniques because we found no evidence that subgroups of patients received only one of the compared treatments or nonuniform treatment effects over the propensity score (39,40). We performed a propensity score analysis to compare treatments among patients of similar risk profiles and found minimal differences between covariate-adjusted and propensity score–adjusted hazard ratios. However, both traditional proportional hazards models and propensity score techniques can only account for known measured confounders (41,42). Neither can overcome the inability to measure critical confounders that influence survival such as functional status. Nevertheless, whereas unmeasured confounding related to treatment selection is a potential problem in all effectiveness cohorts, and the confidence intervals crossed 1.0 in three of five cohorts, the consistency of our findings across diverse cohorts with different potential for bias suggests that oxaliplatin is not only efficacious in the clinical trial setting but also effective.

In summary, our analyses suggest that the benefit of oxaliplatin evident in clinical trials appears to manifest in day-to-day practice. Physicians and patients should be reassured from our findings that oxaliplatin is associated with marginally but consistently superior survival for patients diagnosed before age 75 years in community settings. Future research should examine the comparative effectiveness and safety of oxaliplatin in high-risk subgroups including patients older than 75 years, patients with severe comorbidity, and racial and ethnic minorities. Systematic methods for the study of these groups must be implemented for comparative effectiveness research to successfully address their outcomes.


The Agency for Healthcare Research and Quality (AHRQ), US Department of Health and Human Services (HHS) as part of the Developing Evidence to Inform Decisions about Effectiveness (DEcIDE) program (Contract No. HSA290-2005-0016-I-TO7-WA1, 36-BWH-1 and HHSA290-2005-0040-I-TO4-WA1, 36-UNC). The authors of the report are responsible for its content. Statements in the report should not be construed as endorsement by the AHRQ or the US Department of HHS. The National Cancer Institute (R01CA131847 to D. Schrag, Principle Investigator) funded work on the Medicaid and SEER–Medicare data. The CanCORS study was supported by grants from the National Cancer Institute to the CanCORS Statistical Coordinating Center and Primary Data Collection and Research Centers (U01 CA093344, U01 CA093332, U01 CA093324, U01 CA093348, U01 CA093329, U01 CA01013, U01 CA093326), and by a grant from the Department of Veteran’s Affairs to the Durham VA Medical Center (CRS 02-164) and (HARO 03-438MO-03). The National Institute on Aging (R01AG023178, T. Stürmer, PI), The National Institute of Diabetes and Digestive and Kidney Diseases (2P30DK034987, Robert Sandler, PI), The ACCENT collaboration is funded by (U01 CA25224) The Centers for Disease Control and Prevention (CDC) through the Association of Schools of Public Health (ASPH), (S3888, M. J. Schymura, PI).

Supplementary Material

Supplementary Data:


Comparative effectiveness research using multiple datasets requires the cooperation of a large number of investigators from different groups. This project could not have been possible without the generosity of multiple teams of investigators who agreed to share their data and who were willing to submit their data for scrutiny and comparison with other sources—without stipulation as to how the data would be presented. Without the willingness to share data and provide assistance regarding the nuances of each data source, this study could not have been accomplished. Study authors acknowledge the critical contribution of the following investigative teams and highlight specific individuals within those teams: AHRQ DEcIDE Cancer Consortium, BWH/Dana-Farber members: Li Ling, Kun Chen, Jennifer Wind; UNC members: Richard Goldberg, Joseph Galanko, Anne-Marie Meyer, Janet Freburger, Alice Fortune-Greeley, Sara Wobker, Anne Jackman. The Adjuvant Colon Cancer End Points Group (ACCENT), specifically the following lead investigators of the included studies: Leonard Saltz (CALGB 89803); Eric van Cutsem (PETACC-3); Aimery de Gramont and Thierry Andre (MOSAIC); Michael O’Connell and Greg Yothers (NSABP C07); James Cassidy and Christopher Twelves (X-ACT). In addition, special thanks to Brian Bot and Jeffrey Meyers at the ACCENT statistical coordinating center at the Mayo Clinic. The Cancer Care Outcomes Research & Surveillance Consortium (CanCORS), the entire leadership of the CanCORS consortium, the statistical coordinating center and the publications committee reviewers. The National Comprehensive Cancer Network (NCCN), the Colorectal Cancer Outcomes Database Principal Investigators, statistical analyst Anna Ter Veer and project manager Dana Milne. NYSCR–Medicare and Medicaid, Investigators from the NYS Cancer Registry and NYS Medicaid Program including Patrick Roohan, Francis Boscoe, and Amber Sinclair. SEER–Medicare linked data, this study used the linked SEER–Medicare database. This resource has been made available to the research community through collaborative efforts of the NCI and the Center for Medicare and Medicaid Services (CMS). 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. Contributions from investigators and research staff too numerous to name, but without whom this work could not have been accomplished are also recognized. T. Stürmer's institution, Brigham and Women's Hospital, Harvard Medical School, Division of Preventive Medicine, has received grants from GlaxoSmithKline, Merck, and Sanofi-Aventis, but not for this work.

Funding sources and collaborating agencies were not directly involved with the design, analysis and interpretation, or writing of the article. Final article approval was provided by: AHRQ, the CanCORS publication committee, the NYSCR, the NCCN publication committee, SEER–Medicare.


1. Howlander N, Noone AM, Krapcho M, et al. SEER Cancer Statistics Review, 1975-2008. Bethesda, MD: National Cancer Institute; 2011. Accessed November 3, 2011. Based on November 2010 SEER data submission, posted to the SEER web site.
2. Gunderson LL, Jessup JM, Sargent DJ, Greene FL, Stewart AK. Revised TN categorization for colon cancer based on national survival outcomes data. J Clin Oncol. 2010;28(2):264–271. [PMC free article] [PubMed]
3. Gill S, Loprinzi CL, Sargent DJ, et al. Pooled analysis of fluorouracil-based adjuvant therapy for stage II and III colon cancer: who benefits and by how much? J Clin Oncol. 2004;22(10):1797–1806. [PubMed]
4. NIH Consensus Conference. Adjuvant therapy for patients with colon and rectal cancer. JAMA. 1990;264(11):1444–1450. [PubMed]
5. Andre T, Boni C, Mounedji-Boudiaf L, et al. Oxaliplatin, fluorouracil, and leucovorin as adjuvant treatment for colon cancer. N Engl J Med. 2004;350(23):2343–2351. [PubMed]
6. Andre T, Boni C, Navarro M, et al. Improved overall survival with oxaliplatin, fluorouracil, and leucovorin as adjuvant treatment in stage II or III colon cancer in the MOSAIC trial. J Clin Oncol. 2009;27(19):3109–3116. [PubMed]
7. Kuebler JP, Wieand HS, O’Connell MJ, et al. Oxaliplatin combined with weekly bolus fluorouracil and leucovorin as surgical adjuvant chemotherapy for stage II and III colon cancer: results from NSABP C-07. J Clin Oncol. 2007;25(16):2198–2204. [PubMed]
8. Murthy VH, Krumholz HM, Gross CP. Participation in cancer clinical trials: race-, sex-, and age-based disparities. JAMA. 2004;291(22):2720–2726. [PubMed]
9. 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–IV-18. [PubMed]
10. Surveillance Research Program of the National Cancer Institute. Overview of the SEER Program. Accessed October 20, 2009.
11. Romanus D, Weiser MR, Skibber JM, et al. Concordance with NCCN Colorectal Cancer Guidelines and ASCO/NCCN Quality Measures: an NCCN institutional analysis. J Natl Compr Canc Netw. 2009;7(8):895–904. [PubMed]
12. NCCN Research & Business Resources. NCCN Oncology Outcomes Database. Accessed August 6, 2010.
13. Ayanian JZ, Chrischilles EA, Fletcher RH, et al. Understanding cancer treatment and outcomes: the Cancer Care Outcomes Research and Surveillance Consortium. J Clin Oncol. 2004;22(15):2992–2996. [PubMed]
14. Kahn KL, Adams JL, Weeks JC, et al. Adjuvant chemotherapy use and adverse events among older patients with stage III colon cancer. JAMA. 2010;303(11):1037–1045. [PMC free article] [PubMed]
15. Twelves C, Wong A, Nowacki MP, et al. Capecitabine as adjuvant treatment for stage III colon cancer. N Engl J Med. 2005;352(26):2696–2704. [PubMed]
16. Van Cutsem E, Labianca R, Bodoky G, et al. Randomized phase III trial comparing biweekly infusional fluorouracil/leucovorin alone or with irinotecan in the adjuvant treatment of stage III colon cancer: PETACC-3. J Clin Oncol. 2009;27(19):3117–3125. [PubMed]
17. Saltz LB, Niedzwiecki D, Hollis D, et al. Irinotecan fluorouracil plus leucovorin is not superior to fluorouracil plus leucovorin alone as adjuvant treatment for stage III colon cancer: results of CALGB 89803. J Clin Oncol. 2007;25(23):3456–3461. [PubMed]
18. Alberts SR, Sargent DJ, Smyrk TC, et al. Adjuvant mFOLFOX6 with or without cetuximab in KRAS wild-type patients with resected stage III colon cancer: results from NCCTG Intergroup phase III trial N0147. J Clin Oncol. 2010;28(suppl):18s. Abstract CRA3507.
19. Goldberg RM, Sargent DJ, Thibodeau SN, et al. Adjuvant mFOLFOX6 plus or minus cetuximab in patients with KRAS mutant resected stage III colon cancer: NCCTG Intergroup phase III trial N0147. J Clin Oncol. 2010;28(suppl):18s. Abstract 3508.
20. Allegra CJ, Yothers G, O’Connell MJ, et al. Phase III trial assessing bevacizumab in stages II and III carcinoma of the colon: results of NSABP protocol C-08. J Clin Oncol. 2011;29(1):11–16. [PMC free article] [PubMed]
21. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373–383. [PubMed]
22. Klabunde CN, Potosky AL, Legler JM, Warren JL. Development of a comorbidity index using physician claims data. J Clin Epidemiol. 2000;53(12):1258–1267. [PubMed]
23. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45(6):613–619. [PubMed]
24. Piccirillo JF, Tierney RM, Costas I, Grove L, Spitznagel EL., Jr Prognostic importance of comorbidity in a hospital-based cancer registry. JAMA. 2004;291(20):2441–2447. [PubMed]
25. Bang D, Piccirillo J, Littenberg B, Johnston A. The Adult Comorbidity Evaluation-27 Test—a new comorbidity index for patients with cancer. J Clin Oncol. 2000;19(suppl) Abstract 1701.
26. Sanoff HK, Sargent DJ, Campbell ME, et al. Five-year data and prognostic factor analysis of oxaliplatin and irinotecan combinations for advanced colorectal cancer: N9741. J Clin Oncol. 2008;26(35):5721–5727. [PMC free article] [PubMed]
27. Little R, Rubin D. Statistical Analysis With Missing Data. New York, NY: John Wiley and Sons, Inc; 2002.
28. Brookhart MA, Schneeweiss S, Rothman KJ, Glynn RJ, Avorn J, Sturmer T. Variable selection for propensity score models. Am J Epidemiol. 2006;163(12):1149–1156. [PMC free article] [PubMed]
29. Hines RB, Chatla C, Bumpers HL, et al. Predictive capacity of three comorbidity indices in estimating mortality after surgery for colon cancer. J Clin Oncol. 2009;27(26):4339–4345. [PMC free article] [PubMed]
30. Haller DG, Tabernero J, Maroun J, et al. Capecitabine plus oxaliplatin compared with fluorouracil and folinic acid as adjuvant therapy for stage III colon cancer [published online ahead of print March 7, 2011] J Clin Oncol. 2011;29(11):1465–1471. [PubMed]
31. Sanoff HK, Bleiberg H, Goldberg RM. Managing older patients with colorectal cancer. J Clin Oncol. 2007;25(14):1891–1897. [PubMed]
32. Sargent DJ, Goldberg RM, Jacobson SD, et al. A pooled analysis of adjuvant chemotherapy for resected colon cancer in elderly patients. N Engl J Med. 2001;345(15):1091–1097. [PubMed]
33. Iwashyna TJ, Lamont EB. Effectiveness of adjuvant fluorouracil in clinical practice: a population-based cohort study of elderly patients with stage III colon cancer. J Clin Oncol. 2002;20(19):3992–3998. [PubMed]
34. Sundararajan V, Mitra N, Jacobson JS, Grann VR, Heitjan DF, Neugut AI. Survival associated with 5-fluorouracil-based adjuvant chemotherapy among elderly patients with node-positive colon cancer. Ann Intern Med. 2002;136(5):349–357. [PubMed]
35. Jessup JM, Stewart A, Greene FL, Minsky BD. Adjuvant chemotherapy for stage III colon cancer: implications of race/ethnicity, age, and differentiation. JAMA. 2005;294(21):2703–2711. [PubMed]
36. Goldberg RM, Tabah-Fisch I, Bleiberg H, et al. Pooled analysis of safety and efficacy of oxaliplatin plus fluorouracil/leucovorin administered bimonthly in elderly patients with colorectal cancer. J Clin Oncol. 2006;24(25):4085–4091. [PubMed]
37. Haller D, Cassidy J, Tabernero J, et al. Efficacy findings from a randomized phase III trial of capecitabine plus oxaliplatin verus bolus 5-FU/LV for stage III colon cancer (NO16968): no impact of age on disease-free survival. Proc Gastrointestinal Cancers Symposium. 2010. Abstract 284. Accessed December 6, 2011.
38. McCleary NAJ, Meyerhardt J, Green E, et al. Impact of older age on the efficacy of newer adjuvant therapies in >12,500 patients with stage II/III colon cancer: findings from the ACCENT database. J Clin Oncol. 2009;27(15s) Abstract 4010. [PMC free article] [PubMed]
39. Glynn RJ, Schneeweiss S, Sturmer T. Indications for propensity scores and review of their use in pharmacoepidemiology. Basic Clin Pharmacol Toxicol. 2006;98(3):253–259. [PMC free article] [PubMed]
40. Sturmer T, Rothman KJ, Glynn RJ. Insights into different results from different causal contrasts in the presence of effect-measure modification. Pharmacoepidemiol Drug Saf. 2006;15(10):698–709. [PMC free article] [PubMed]
41. Sturmer T, Schneeweiss S, Avorn J, Glynn RJ. Adjusting effect estimates for unmeasured confounding with validation data using propensity score calibration. Am J Epidemiol. 2005;162(3):279–289. [PMC free article] [PubMed]
42. Sturmer T, Joshi M, Glynn RJ, Avorn J, Rothman KJ, Schneeweiss S. A review of the application of propensity score methods yielded increasing use, advantages in specific settings, but not substantially different estimates compared with conventional multivariable methods. J Clin Epidemiol. 2006;59(5):437–447. [PMC free article] [PubMed]

Articles from JNCI Journal of the National Cancer Institute are provided here courtesy of Oxford University Press