Presented in part at the Canadian Association for Health Services and Policy Research (CAHSPR) Conference, Toronto, Ontario, May 10-13, 2009.
Many patients with stage III colon adenocarcinoma do not receive adjuvant chemotherapy despite the proven survival advantage it offers. To enhance the provision of optimal cancer care, patient characteristics associated with not receiving guideline-adherent treatment must be identified among patients with operable, stage III colon adenocarcinoma.
This was a population-based, retrospective study of all patients who underwent surgery for stage III colon adenocarcinoma diagnosed from 2002 through 2005 in Alberta, Canada. Demographic and treatment information captured in the Alberta Cancer Registry were linked to: 1) hospital discharge data to determine comorbidities, 2) electronic medical records to identify consults with oncologists, and 3) the 2001 Canadian census for neighborhood-level socioeconomic data. Multivariate log-binomial regression models were used to identify patient characteristics that were associated with not having a consultation with a medical oncologist and not receiving adjuvant chemotherapy.
Of the 772 patients who underwent surgery for stage III colon adenocarcinoma and met the eligibility criteria, 618 patients (80%) had a consultation with an oncologist. Of those, 388 patients (63%) initiated adjuvant chemotherapy within 84 days of their surgery. Patient characteristics that were associated with not having a consultation with an oncologist were neighborhood income, geography, age, and comorbidities. Of those patients who had a consultation, after adjusting for comorbidities, only older age was related to not receiving adjuvant chemotherapy.
Treatment guidelines for stage III colon adenocarcinoma have been in place since 1990,1 after a large, randomized, controlled trial2 demonstrated a relative risk reduction of 40% for recurrence and of 33% for mortality in patients who received postsurgical (ie, adjuvant) chemotherapy. The guidelines state that patients with stage III adenocarcinoma should undergo surgery followed by chemotherapy. Despite the survival advantage and the general acceptance of adjuvant chemotherapy as the standard of care, several studies conducted in the United States have indicated that a large proportion of patients do not receive it or experience treatment delays.3-8 To our knowledge, there has only been 1 study conducted in Canada that has measured adherence to guidelines for patients with stage III colon adenocarcinoma.9 That population-based study, which covered 3 provinces, indicated that approximately 50% of patients who were diagnosed in 2004 did not receive adjuvant chemotherapy.
To deliver optimal, evidence-based cancer care, it is important that cancer treatment guidelines are followed. In the current population-based study, we investigated patient characteristics associated with not receiving care according to the guidelines for patients who are diagnosed with stage III colon adenocarcinoma in Alberta. In Alberta, all nonsurgical cancer treatment occurs in provincially coordinated and accredited cancer care facilities. To receive treatment according to the guidelines, a patient must undergo surgery, be referred to a cancer care facility for a consultation with an oncologist, attend the consult, be recommended to receive chemotherapy by the oncologist, and then agree to receive the chemotherapy. We investigated patient characteristics associated with not having a consultation with an oncologist and not receiving adjuvant chemotherapy as a step toward understanding barriers to receiving guideline-recommended treatment for patients with stage III colon adenocarcinoma. Factors of interest included patient and disease characteristics, region of residence, and several neighborhood-level socioeconomic variables based on census data.
MATERIALS AND METHODS
In Alberta, a province with a population of approximately 3 million, physicians and hospitals are legally required to report all cancer cases to the Alberta Cancer Registry, a member of the North American Association of Comprehensive Cancer Registries that has a proven record of high quality in its annual review.10 By using this comprehensive registry, all patients who had a histologically confirmed diagnosis of stage III colon adenocarcinoma (International Classification of Diseases for Oncology [ICD-O],11 code c18) between 2002 and 2005 and underwent surgery were identified and included in the study. Cancer stage was determined using the American Joint Committee on Cancer (AJCC) (6th edition)12 staging rules. Stage III is defined as a tumor that has spread to regional lymph nodes (N1 or N2) but not to distant metastatic sites (M0). Patients were excluded if they died within 1 week of their diagnosis (to eliminate those who did not have an opportunity to receive appropriate treatment), if they were diagnosed with another cancer within the preceding or subsequent 6 months of their diagnosis (to avoid treatment decisions that might have been influenced by another cancer), or if they were treated outside of Alberta.
Extraction/Collection of Data on Potential Explanatory Variables
Histologic grade (the sixth digit of the ICD-O code, third edition11), sex, age at diagnosis, postal code at diagnosis, and healthcare region (termed Regional Health Authority [RHA]) of residence at diagnosis were obtained from the Alberta Cancer Registry. During the period in question (2002-2005), there were 17 cancer facilities throughout the province, 6 of which had oncologists (or, at 1 facility, internists with an interest in medical oncology) on site to provide treatment consultations with patients, including 2 tertiary cancer centers and 4 associate cancer centers (Fig. 1). Oncologists located at these 6 cancer facilities provided consultations to discuss treatment options for all cancer patients in the province. The 2 tertiary centers are located in the largest cities, Edmonton (RHA 6) and Calgary (RHA 3), where 66% of the provincial population resides. The 4 associate cancers centers are located in smaller cities located throughout the province and include 2 in southern Alberta, 1 in central Alberta, and 1 in northwestern Alberta. The minimum distance between each patient's residential postal code at the time of their diagnosis and the postal code of the closest of the 6 cancer facilities that provide consults with oncologists was calculated using the driving direction link from the internet Yellow Pages (available at: http://www.yellowpages.ca accessed June 19, 2009). We assumed that all patients living in the 2 large cities of Alberta, Edmonton and Calgary, had the same driving distance to the respective city's cancer care facility, which was calculated as the average distance for a 15% random sample of residential postal codes within each city's boundaries.13
Data from the 2001 Canadian census were used to obtain area-level variables that were related to socioeconomic status in which the geographic area, called a dissemination area, is a neighborhood of approximately 600 households. Five variables that were indicative of socioeconomic status14 were included: 1) median income; 2) proportion of employment; 3) proportion not graduated from high school; 4) proportion separated, divorced, or widowed; and 5) proportion living alone.
Comorbidity scores were calculated using outpatient data from all Alberta hospitals, known as the Ambulatory Care Classification System, and inpatient data from all Alberta hospitals, known as the Discharge Abstracts Database. Specifically, diagnosis codes from these databases were obtained for patients in the cohort that occurred within 1 year before diagnosis and were merged with the cohort dataset by using patient identifiers; International Classification of Diseases version 9-Clinical Modification (comorbidities) (ICD-9-CM) and International Classification of Diseases version 10-Canada (ICD-10-CA) diagnosis codes were used to calculate a comorbidity score for each patient. Comorbidity scores were calculated based on an updated version of the index described by Deyo et al.15 and developed by Quan et al.16 using enhanced ICD-9-CM coding algorithms and incorporating ICD-10 codes. For the purposes of analysis, these scores were categorized as 0 and ≥1, which represent “no serious comorbidity” and “1 or more serious comorbidities,” respectively.
Extraction/Collection of Outcome Data
The primary outcomes of interest were: 1) not having a consultation with an oncologist within 6 months of diagnosis and 2) not receiving standard treatment, which was defined as surgery plus adjuvant chemotherapy within 84 days postsurgery (the outer time limit for initiating adjuvant chemotherapy specified in the colon cancer clinical practice guidelines for Alberta).17 Previously, we observed that consultations with oncologists that occurred more than 6 months after the diagnosis date were not related to initial treatment discussions18; therefore, we limited the first outcome measure to consults within 6 months of diagnosis.
To define the outcomes, starting dates and modalities of all initial treatments received for curative intent (including surgery) were obtained from the Alberta Cancer Registry. If the dates of adjuvant chemotherapy were missing or incomplete, then the cancer electronic medical record was reviewed to complete the missing date information. The cancer electronic medical record was implemented in 2000/2001 and has been used extensively with comprehensive quality-assurance procedures in place. The date and facility of the first postsurgical consultation with an oncologist also was obtained from the electronic cancer medical record.
The analysis was conducted in 2 stages. In the first stage, associations of the potential explanatory variables with failing to have a consultation were assessed in the entire eligible patient cohort. Because having a consult with an oncologist is a required precursor to receiving adjuvant chemotherapy, the second stage of the analysis examined associations of the potential explanatory variables with failing to receive adjuvant treatment in the subcohort of patients who had a consultation with an oncologist.
In each stage of the analysis, we started with descriptive analysis, quantifying the proportion who did not receive a consultation with an oncologist and, among those who attended the consultation, the proportion who did not receive adjuvant treatment. Then, we calculated these proportions stratified by each of the potential explanatory variables, assessing their bivariate associations using chi-square tests (or Fisher exact tests if necessary). To summarize the independent associations of multiple explanatory variables simultaneously, log-binomial regression analysis with backward selection was conducted. For convergence issues in the log-binomial model, the COPY method19 was used with 999 copies. To facilitate interpretation, this method, which provides relative risk estimates, was selected for use in the current analysis over the more popular logistic regression, which provides odds ratio estimates. P values and 95% confidence intervals were calculated using a large-sample approximation of the regression.
Continuous variables were evaluated both as continuous and categorical variables as part of exploratory data analysis. Cutoff points for categorizing continuous variables in the final analyses were based on identified thresholds beyond which an effect was observed. Neighborhood median income, however, was categorized into quartiles based on the entire Alberta population. Histologic grade was categorized as well/moderately differentiated (grades 1 and 2) and poorly/undifferentiated (grades 3 and 4). Lymph node status was categorized as 1 to 3 positive lymph nodes and ≥4 or more positive lymph nodes, corresponding to lymph node status of N1 and N2, respectively, using AJCC TNM classification rules.12 For analyses that involved the region of residence, some regions were collapsed because of small numbers; specifically, the following RHAs were combined: RHAs 1 and 2 were combined as Southern Alberta; RHAs 4 and 5 were combined as Central Alberta; and RHAs 7, 8, and 9 were combined as Northern Alberta. Age was categorized as <70 years, 70 to 74 years, and >74 years to maintain reasonable numbers in each age category and to ensure consistency across analyses. All analyses were performed using SAS statistical software version 9.1 (SAS Institute, Inc., Cary, NC).
In total, 842 Alberta residents were diagnosed with stage III colon cancer between 2002 and 2005. Sixty-nine patients (8.2%) were excluded for 1 or more of the following reasons: 17 patients did not have an adenocarcinoma, 13 patients died within 7 days of their diagnosis, 3 cases were not histologically confirmed, 41 patients had another cancer diagnosis within 6 months, 3 patients did not undergo surgery, and 1 patient treated outside of Alberta. Our analysis included the remaining 772 patients with stage III, surgically treated colon adenocarcinoma.
Tables 1 and 2 display the distribution of various patient characteristics and socioeconomic indicators that were evaluated in the current study, respectively, and the relation of those characteristics to each of the outcome variables. Of the 772 patients, 618 (80%) had a consultation with an oncologist within 6 months of their diagnosis. Of those 618 patients, 388 (63%) received adjuvant chemotherapy.
Table 1. Distribution of Demographic Characteristics of Patients With Stage III Colon Adenocarcinoma With Respect to Undergoing Surgery, Not Having a Consult With an Oncologist Within 6 Months of Diagnosis, and Not Receiving Adjuvant Chemotherapy Within 84 Days of Surgery
Percentages are row percentages; denominator for percentage is the number of patients who underwent surgery.
Guideline treatment is surgery plus adjuvant chemotherapy within 84 days of surgery.
Percentages are row percentages; denominator for percentage is the number of patients who had a consult with an oncologist.
Patients with stage III colon cancer
Age at diagnosis, y
Year of diagnosis
Region of residence at diagnosis
Edmonton and area
Calgary and area
Histologic grade of differentiation
No. of positive lymph nodes
Distance to closest cancer facility, km
Table 2. Distribution of Neighborhood-Level Socioeconomic Characteristics of Patients With Stage III Colon Cancer With Respect to Undergoing Surgery, Not Having a Consult With an Oncologist Within 6 Months of Diagnosis, and Not Receiving Adjuvant Chemotherapy Within 84 Days of Surgery
Table 3 presents the results of the multivariate log-binomial regression for both outcome variables. Factors that were associated independently with not receiving a consultation were the presence of comorbidities (P = .0002), older age (P < .0001), residence outside of the Edmonton area (RHA 6; P = .029), and lower median neighborhood income (P = .001). Patients aged ≥75 years were 8.7 times more likely not to have a consultation with an oncologist than patients aged <70 years. Compared with patients who lived in the Edmonton area, those who lived in Southern Alberta and those who lived in the rest of the province were 1.7 times and approximately 1.4 times more likely to not have a consultation with an oncologist, respectively. Patients who lived in the lowest median income quartile neighborhoods were twice as likely to not have a consultation with an oncologist compared with patients who lived in the highest median income quartile neighborhoods. Of those patients who had a consultation with an oncologist, those aged ≥75 years were approximately 3 times more likely to not receive adjuvant chemotherapy than those aged <70 years after adjusting for the presence of serious comorbidities.
Table 3. Adjusted Relative Risks and 95% Confidence Intervals for Factors Related to Not Having a Consult With an Oncologist Within 6 Months of Surgery and Not Receiving Adjuvant Chemotherapy Within 84 Days of Surgerya
Adjusted RR (95% CI)
Not Having a Consult
Not Receiving Guideline Treatment
CI indicates confidence interval; RR; relative risk; Q, quartile.
Adjusted for all variables listed in the table for each outcome variable.
After a series of studies conducted in the United States that identified patient characteristics associated with not receiving guideline-recommended adjuvant chemotherapy in patients with stage III colon adenocarcinoma,3-9 in the current study, we investigated whether similar population-based observations could be made in a similar patient cohort in Alberta. Patient characteristics that were associated with not having a consultation with an oncologist also were investigated.
Between 2002 and 2005, 20% of patients with stage III colon adenocarcinoma in Alberta did not have a consultation with an oncologist, which is a prerequisite to receiving adjuvant chemotherapy. The variability of patients who had a consultation with an oncologist based on region and income is concerning, especially because consulting with a specialist is a service that does not have any direct cost to any resident of Alberta. It is possible, however, that residents in lower income areas are less able to take time off from work or have other barriers, such as transportation or child care issues, to attending a consultation with an oncologist. It is noteworthy that, the regional variation was not explained by distance to the closest cancer facility, although many patients traveled >200 kilometers to consult with an oncologist. This finding is consistent with other studies that evaluated treatment in patients with lung cancer20 and the receipt of adjuvant chemotherapy in patients with colorectal cancer21 living in rural regions. Although it was not possible in the current study to ascertain reasons for the regional differences observed, variation in referral patterns by regional surgeons may explain them (if a patient with cancer undergoes surgery, then it is the surgeon, not the primary care physician, who refers the patient to the oncologist); this possibility needs to be better elucidated.
Few studies have quantified the percentage of patients with stage III colon adenocarcinoma who had a consultation with an oncologist. However, 1 large study conducted in the United States that included patients who were insured under Medicare (patients aged ≥65 years) indicated that 78% of patients with stage III colon adenocarcinoma had a consultation with a medical oncologist within 6 months of their diagnosis,22 consistent with our finding of 80%. In that study, factors that were related to consulting with an oncologist included age, race, sex, marital status, year of diagnosis (1992-1999), region of residence, tumor grade, number of positive lymph nodes, comorbidities, and poverty level,22 similar to our current findings. However, in our study, sex, marital status, year of diagnosis (2002-2005), and tumor grade were not related to whether a patient had a consultation with an oncologist, and we were not able to assess race or the number of positive lymph nodes.
Overall, 50% of patients in the current study did not receive timely adjuvant chemotherapy. Consistent with other studies conducted in the United States,3-8 age was the factor related most significantly to not receiving guideline-recommended treatment after adjusting for comorbidities, although studies have demonstrated that neither the survival advantage6, 23 nor toxicity23 of adjuvant chemotherapy are related to age in patients with stage III colon adenocarcinoma.
Large studies conducted in the United States based on Medicare patients (aged ≥65 years) diagnosed in the early to middle 1990s indicated that 50%8 to 55%5 of patients with stage III colon adenocarcinoma received adjuvant chemotherapy, whereas an estimate for patients who were diagnosed in 20023 was 63%. A study that was conducted on patients identified from 3 regions that comprise part of the California Cancer Registry in 1996 and 19977 indicated that 67% of patients with stage III colon adenocarcinoma who underwent surgery also received adjuvant chemotherapy. Our finding that, overall, 50% of patients with stage III colon adenocarcinoma underwent surgery was lower compared with findings from the US studies described above. This difference may be attributable to differences in the private and public healthcare systems. Specifically, uninsured patients in the United States may not be captured in these nonpopulation-based studies; whereas, in Alberta, everyone is insured publicly, and the complete cost of adjuvant chemotherapy for stage III colon cancer is covered by the provincial healthcare insurance plan.
A limitation to the current study is that we did not have detailed data related to surgical complications that may have delayed or prevented wound healing or patient functional status24; therefore, we do not know definitively which patients were not medically appropriate candidates for chemotherapy. Studies conducted in the United States3-8 have had the same limitations; thus, comparisons can be made across these studies with respect to both the percentages of patients who did not receive adjuvant chemotherapy and the factors related to not receiving adjuvant chemotherapy.
Functional status is a potentially important clinical factor for recommending adjuvant chemotherapy and, although it is related to the presence of comorbidities, reportedly is poorly correlated with comorbidity scores.24 Future studies should capture both functional status and comorbidities to determine whether they explain most of the age effect for not receiving adjuvant chemotherapy among patients with stage III colon adenocarcinoma.
Another study limitation is that we were not able to identify whether or not patients refused chemotherapy. Other studies have demonstrated that physician recommendation is the largest factor for determining whether a patient accepts a particular treatment option.25, 26 It has been reported that, to accept a certain treatment, older patients require a greater survival advantage than younger patients for the same toxicity risk.26, 27 The adjuvant chemotherapy options for patients with stage III colon cancer, however, are fairly well tolerated and do not usually predispose patients to high-grade toxicities. Patient preference, therefore, is not likely to explain the large age effect we observed in the current study.
In summary, we observed that, overall, 50% of patients with stage III colon adenocarcinoma did not receive guideline-recommended adjuvant chemotherapy in Alberta between 2002 and 2005. Approximately half of those patients did not receive adjuvant treatment because they did not have a consultation with an oncologist. In turn, factors that were associated independently with not having a consultation included socioeconomic indicators and geography, indicating a fixable system problem that needs to be addressed.
We thank Charlotte King for assistance in obtaining and preparing the data for analysis and Angela Bella for assistance in the formatting of this article
CONFLICT OF INTEREST DISCLOSURES
This research was made possible by grants received from the Canadian Institute for Health Research, the Canadian Cancer Society, and the Alberta Cancer Foundation.