• acute care indicators;
  • cancer survivors;
  • chronic comorbidities;
  • colorectal carcinoma;
  • quality of care


  1. Top of page
  2. Abstract
  6. Acknowledgements

Comorbid conditions are the major threat to life for many cancer survivors, yet little is known about the quality of the noncancer-related health care they receive. The authors analyzed the Medicare claims of 14,884 Medicare-eligible, 5-year colorectal carcinoma survivors who were diagnosed initially while they lived in a region monitored by the Surveillance, Epidemiology, and End Results (SEER) Program and compared them with matched controls who had no history of cancer drawn from the Medicare 5% sample. In both univariable and multivariable analyses, cancer survivorship was associated with an increased likelihood of not receiving recommended care across a broad range of chronic medical conditions (odds ratio, 1.19, 95% confidence interval, 1.12–1.27). For example, colorectal carcinoma survivors were less likely than controls to receive appropriate follow-up for heart failure, necessary diabetic care, or recommended preventive services. Having both primary care physicians and oncologists involved in follow-up appeared to ameliorate this effect significantly. African-American, poor, and elderly patients were less likely to receive necessary care in both groups. Whether it was due to patient factors, physician factors, or both, cancer survivors appear to be a vulnerable patient population, because their cancer diagnosis may shift attention away from important noncancer problems and providers. In addition, there may be lack of clarity around the relative roles primary care and specialist physicians will play in a survivor's care. Special attention and education are needed to ensure that survivors receive optimal medical services. Cancer 2004. © 2004 American Cancer Society.

There are over 8 million people living in the United States with a personal history of cancer,1 and this segment of the population is expected to continue to grow. In fact, most patients who are diagnosed with cancer today will not die from it.2 Therefore, preventive care and treatment of comorbid conditions remain important for these patients despite their cancer diagnosis.

Cancer survivors, although they may be fortunate in many ways, may represent a vulnerable population of patients. We reported previously that breast carcinoma survivors received routine preventive services more frequently than matched controls.3 However, this was due largely to their more frequent contacts with more physicians. In fact, those who were cared for only by primary care physicians actually were less likely to undergo recommended cancer screening and surveillance. In addition, previous studies have shown that patients from vulnerable populations have difficulty accessing physicians, are less likely to receive preventive health maintenance, and are more likely to have emergency room visits; and these factors are associated with poor health outcomes.4, 5

Asch et al. demonstrated that under use can be evaluated among Medicare beneficiaries with a broad series of valid, evidence-based measures using administrative claims data.6 Here, we have applied these methods to a population-based cohort of colorectal carcinoma survivors to compare the quality of noncancer care they received with that experienced by matched controls without cancer and to examine the effect of physician specialty on the delivery of that care.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Data Sources

Two data sources were used: 1) data from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) registry; and 2) Centers for Medicare and Medicaid Services (formerly the Health Care Financing Administration [HCFA]) Medicare claims data. The 11 tumor registries participating in the SEER Program capture ≈ 97% of their incident cases,7 covering a representative sample8 of approximately 14% of the United States population.9 Registries collect data on each patient's age, gender, race and ethnicity, cancer site, disease stage, histology, date of diagnosis, and the date and cause of death. The SEER registries also record initial treatment data on surgery and radiation received in the first 4 months after diagnosis. However, SEER does not provide information on later treatments. To follow these patients longitudinally, Medicare claims for patients age ≥ 65 years have been linked to the SEER registry data. The Medicare data base includes files through 1998 for inpatient and outpatient care, physician and laboratory billings, as well as bills for home health and hospice care. For patients age ≥ 65 years who are captured by the SEER registries, 94% have been linked to Medicare.10 Census-level sociodemographic data also have been linked to these patients.

Cohort Selection

We chose to study colorectal carcinoma, because it is a common disease with relatively high survival rates. In addition, survivors of colorectal carcinoma generally do not have severe long-term consequences from the disease or its treatment. Consequently, the effects of survivorship itself may be isolated relatively from disease-related sequelae.

Cases consisted of all patients who were Medicare-eligible on the basis of age, with a past history of colorectal carcinoma that was diagnosed while the patient resided in 1 of the 11 SEER regions at least 5 years prior to the time of analysis. Patients were excluded if they were enrolled in a Health Maintenance Organization (HMO) at any time during the study period or if they were not eligible for both parts of Medicare. We defined colorectal carcinoma survivors as patients who were 1) diagnosed with invasive colorectal carcinoma in 1991 or 1992, 2) had nonmetastatic disease at diagnosis, 3) survived through the end of 1998, 4) had not been diagnosed with subsequent malignancies recorded by SEER, 5) had not received chemotherapy or radiation in 1997–1998, 6) had no diagnostic codes for metastatic carcinoma in any 2 bills separated by 30 days, and 7) had not been enrolled in hospice. Controls were obtained from a 5% Medicare random sample of patients with no history of cancer and then were matched to each case based on age, gender, race, and geographic location.

Quality-of-Care Measures

The RAND group previously developed and operationalized a comprehensive set of indicators of under use of necessary care using administrative data.6 We used the technical document from the RAND project to calculate the same indicators using our administrative data. This system had been tested and validated on earlier claims before their application by RAND. Additional quality-of-care indicators were selected from the Health Plan Employer Data and Information Set, focusing on conditions that are important health threats that feasibly could be assessed using administrative data, such as influenza vaccination and lipid testing and, in women, cervical examination and bone densitometry. Because we expected that survivorship may have more effect on the management of ongoing chronic comorbid conditions than on acute care, we divided the indicators into these two groups using a cut off of whether they were to take place within 3 months of an index event and analyzed them both separately and together.

Explanatory Variables

We collapsed race and ethnicity into white, African-American, and other categories. Region-specific socioeconomic quintiles were developed based on the availability of information, according to the following hierarchy: 1) race-specific and age-specific median household income by census tract, 2) unadjusted median household income by census tract, 3) median household wealth, and 4) per capita income.11 We identified comorbidities by looking for diagnostic billing codes for various conditions during the study period, using the Deyo implementation12 of the Charlson score.13 The use of chemotherapy and radiation was identified from billing claims, as we have done previously.14 We considered that a patient received care in a teaching hospital if there was a bill for indirect medical education at one of their medical contacts during the study period. To assess the type of providers patients were seeing, we categorized physicians based on their specialty identification in Medicare. Oncology specialists were defined as subspecialists in Medical Oncology or Hematology-Oncology (HCFA specialty codes 83 or 90, respectively), Surgeons (code 02, General Surgery; code 28, Colorectal Surgery; codes 49 and 91, Surgical Oncology), or Radiation Oncologists (code 92). Primary care physicians were identified with the following provider codes for Generalist Physicians: 01, General Practice; 11, Internal Medicine; 08, Family Practice; 16, Obstetrics/Gynecology; 38, Geriatric Medicine; and 70, Multispecialty Group Practices.

Statistical Analyses

We compared the noncancer-related health services received by survivors with those received by controls in the 2-year period from 1997 to 1998. For each indicator, we included only beneficiaries who were eligible to receive the necessary care. Univariate analyses comparing survivors with matched controls were carried out using t tests for continuous variables and chi-square analyses for categorical variables. Sensitivity analyses restricted the cohort to patients age ≤ 79 years (the median age) included those survivors who originally presented with in situ disease and excluded patients and controls with any bill containing a cancer diagnosis, even though it may have been referring to their prior cancer diagnosis. Analyses also were carried out that were restricted only to the indicators based on either randomized controlled trials or high-quality observational studies.6 The results were similar for all of these secondary analyses; therefore, they are not reported here.

To facilitate multivariable analyses, we also assessed a composite endpoint defined as the number of necessary services received or avoidable outcomes not experienced divided by the number of indicators for which the patient was eligible. Variables that correlated with this continuous score on univariable analyses were then entered into a multivariable linear regression model to predict the composite score and into logistic regression models to predict scores below the 50th and 25th percentiles. Forward stepwise elimination was used. Each of these methods identified similar predictive characteristics; therefore, for clarity, only the logistic regression model that predicted patients with scores below the 25th percentile are reported here. Interaction terms between the significant explanatory variables were investigated further. Separate models were created for acute care indicators, chronic care indicators, and a summary of all indicators. All statistical analyses were performed with Statistical Analysis Software (SAS; version 8.1 for Windows, SAS Institute Inc., Cary NC).


  1. Top of page
  2. Abstract
  6. Acknowledgements

In total, 18,699 patients with nonmetastatic colorectal carcinoma met the initial eligibility criteria of surviving and participating continuously in both parts of Medicare through the end of 1998. These patients were matched one-to-one with controls. After excluding patients who had evidence of chemotherapy or radiation use, received hospice services, or had a diagnosis of metastatic carcinoma on their bills during the study period, there were 14,884 survivors and 16,659 controls. Among colorectal carcinoma survivors, 35.9% had had locally advanced disease (T4 or lymph node-positive) at the time of their original diagnosis, and 81.6% had lymph node-negative disease. Table 1 summarizes the characteristics of the study sample. The population was quite elderly, with an average age of just under 80 years for both groups. The cohorts were well matched. Survivors were more likely to be diabetic and to have chronic heart and lung disease, whereas dementia was more common among the control group.

Table 1. Demographic Characteristics of the Study Population
CharacteristicSurvivors (n = 14,884)Controls (n = 16,659)
  • CPD: chronic pulmonary disease; CHF: congestive heart failure.

  • a

    P < 0.01 for univariable comparisons with survivors.

Mean age (yrs)79.979.8
Male gender (%)42.443.3
Race (%)  
 African American5.45.4
Urban setting (%)80.575.5a
Teaching hospital (%)20.616.3a
Comorbidities (%)48.744.4a
Diabetes (%)16.914.8a
CPD (%)14.713.2a
CHF (%)15.212.5a
Dementia (%)4.14.8a

Receipt of Necessary Care

Table 2 shows results for the management of chronic comorbid conditions. In almost all situations, colorectal carcinoma survivors were less likely to receive the recommended care. For example, survivors with angina, congestive heart failure (CHF), or chronic lung disease were less likely to have recommended routine visits every 6 months. Only 64.3% of survivors with chronic stable angina had monitoring of their lipid profiles, compared with 69.1% of controls. Survivors with diabetes were significantly less likely than controls to have regular follow-up visits and yearly vision examinations, and there was a trend toward less intensive monitoring of glycosylated hemoglobin and fructosamine. Survivors also appeared to be more likely to experience some avoidable outcomes. For example, survivors with chronic lung disease were more likely than controls to require admission to hospital for this condition. In addition, survivors were less likely to undergo routine preventive care, such as eye examinations, influenza vaccination, or cholesterol screening. Female survivors were less likely to have cervical screening or bone densitometry.

Table 2. Proportion of Patients Receiving Recommended Management of Chronic Comorbidities and Preventive Care
VariableControls (%)Survivors (%)No.P value
  1. NS: nonsignificant; COPD: chronic obstructive pulmonary disease.

 One visit every 6 mos for patients with chronic stable angina96.394.12797< 0.01
 Visit every 6 mos for patients with congestive heart failure94.187.56321< 0.001
 Visit every 6 mos for patients with COPD93.390.65347< 0.001
 Visit every yr for patients with diagnosis of transient ischemic attack97.496.21284NS
 Cholesterol test every 6 mos for patients hospitalized for acute myocardial infarction and who have hypercholesterolemia38.439.8261NS
 Lipid profile ≤ 1 yr after initial diagnosis of angina69.164.323610.01
 Visit every 6 mos for patients with diabetes94.593.262430.03
 Eye examination every yr for patients with diabetes30.127.154210.01
 Glycosylated hemoglobin or fructosamine every 6 mos for patients with diabetes25.523.762430.09
 Influenza vaccination55.453.231,543< 0.001
 Cholesterol screening39.436.531,543< 0.001
 Assessment of visual impairment every 2 yrs50.647.431,543< 0.001
 Mammography in female patients age < 76 yrs51.554.051140.08
 Cervical screening in female patients21.917.818,018< 0.001
 Bone densitometry in female patients5.74.218,018< 0.001
Avoidable outcomes    
 Among patients with known angina, ≥ 3 emergency department visits for cardiovascular-related diagnoses in 1 yr2.22.78187NS
 Among patients with known cholelithiasis, diagnosis of perforated gallbladder0.10.14437NS
 Among patients with known emphysema, subsequent admission for a respiratory diagnosis29.235.52786< 0.001
 Among patients with known COPD, subsequent admission for a respiratory diagnosis18.522.511,370< 0.001
 Nonelective admission for congestive heart failure12.012.27526NS
 Among patients with known diabetes, admission for hyperosmolar or ketotic coma0.10.29701NS
 Among patients with pneumonia, diagnosis of lung abscess or emphysema0.40.59982NS

Table 3 lists acute care indicators. Here, there were few systematic differences in the care received by survivors and controls. Survivors were less likely to have the recommended follow-up visit within 4 weeks after they were discharged from hospital after an acute myocardial infarction or a cerebrovascular accident, with a similar trend after admissions for CHF, but they were more likely to have recommended chest X-rays and electrocardiograms after a diagnosis of CHF. Cancer survivors also were more likely to have at least 1 physician visit each year (89% vs. 86%; P < 0.001).

Table 3. Proportion of Patients Receiving Necessary Acute Interventions
IndicatorControls (%)Survivors (%)No.P value
  1. NS: nonsignificant.

Visit ≤ 4 weeks after discharge of patients hospitalized for acute myocardial infarction75.567.86630.03
Electrocardiogram during emergency department visit for unstable angina94.995.2621NS
Follow-up visit or hospitalization ≤ 1 wk after initial diagnosis of unstable angina36.838.11450NS
Visit ≤ 4 wks after discharge for patients hospitalized for unstable angina79.283.7693NS
Visit ≤ 4 wks after discharge of patients hospitalized for congestive heart failure76.273.028160.06
Electrocardiogram ≤ 3 mos after initial diagnosis of congestive heart failure54.757.768410.01
Chest radiograph ≤ 3 mos after initial diagnosis of congestive heart failure53.561.06841< 0.001
Visit ≤ 4 wks after discharge of patients hospitalized for cerebrovascular accident72.565.51366< 0.01
For patients hospitalized for carotid artery stroke, carotid imaging ≤ 2 wks of initial diagnosis55.050.4314NS
For cerebrovascular accident patients with eventual carotid endarterectomy, interval between carotid imaging and carotid endarterectomy < 2 mos62.972.5166NS
Electrocardiogram within 2 days of initial diagnosis of transient ischemic attack29.329.62124NS
Visit ≤ 4 wks after discharge for patients hospitalized for transient ischemic attack80.078.6353NS
For transient ischemic attack patients with eventual carotid endarterectomy, interval between carotid imaging and carotid endarterectomy < 2 mos69.660.969NS
Visit ≤ 4 wks after discharge of patients hospitalized for diabetes74.074.22047NS
Visit ≤ 2 wks after discharge of patients hospitalized for depression57.260.1669NS
Visit ≤ 4 wks after discharge for patients hospitalized with malignant or otherwise severe hypertension72.275.695NS
Visit ≤ 4 wks after discharge for patients hospitalized for gastrointestinal bleeding74.469.9829NS
Cholecystectomy (open or laparoscopic) for patients with cholelithiasis and ≥ 1 of the following: cholecystitis, cholangitis, gallstone pancreatitis36.332.5938NS
Arthroplasty or internal fixation of hip during hospital stay for hip fracture70.675.9699NS

Effect of Provider Type on Preventive Services for Survivors

Fifty percent of the survivors (7465 patients) continued to see an oncologist in follow-up, and 8% of those survivors (587 patients) saw only an oncologist (Table 4). In all categories of care, patients who were followed by both oncologists and primary care physicians received the highest proportion of recommended care, followed by patients who were followed by primary care physicians. Patients who were followed only by oncologists received significantly worse preventive care compared with patients who also had a primary care physician. Survivors who did not receive care from an oncologist were less likely to undergo cancer-related procedures of surveillance colonoscopy (27.6% vs. 46.7%) and mammography (26.5% vs. 31.3%) compared with patients who saw an oncologist. Conversely, the subset of patients who were seen only by primary care physicians were more likely to receive influenza vaccination (55.2% vs. 43.6%), cervical screening (14.7% vs. 8.2%), and bone densitometry (3.9% vs. 1.1%) compared with patients who were followed only by an oncologist.

Table 4. Correlation between the Type of Provider and the Proportion of Recommended Care Received by Survivors
ProvideraNo. of patientsOdds ratio (95% CI)
OverallAcute interventionsManagement of comorbidityPreventive care
  • 95% CI: 95% confidence interval; PCP: primary care physician.

  • a

    “Both” indicates that patients saw both a primary care physician and an oncology specialist (compared with patients who saw neither); “PCP” indicates that patients saw a primary care physician only; “Oncologist,” patients saw an oncologist only; and “Neither” indicates that patients saw neither a primary care physician nor an oncologist in follow-up.

Both68780.644 (0.639–0.648)0.314 (0.304–0.324)0.720 (0.712–0.729)0.469 (0.462–0.476)
PCP56330.572 (0.566–0.577)0.245 (0.235–0.255)0.708 (0.698–0.718)0.365 (0.357–0.372)
Oncologist5870.534 (0.516–0.552)0.108 (0.085–0.130)0.637 (0.602–0.673)0.353 (0.329–0.377)
Neither17860.283 (0.273–0.293)0.027 (0.020–0.034)0.692 (0.672–0.713)0.080 (0.071–0.088)

Multivariable Analyses

Multivariable analyses on a number of composite endpoints confirmed that being a colorectal carcinoma survivor was associated independently with lower rates of receipt of necessary care, even when adjusting for factors such as age, race, socioeconomic status (SES), and comorbidity. Table 5 shows the effects of these variables on the odds of being in the lowest quartile on the composite measure of the proportion of necessary care received. Survivors had an adjusted odds ratio of 1.19 that they would be in the group that received less necessary care. The only stronger patient characteristic that predicted poor care was African-American race. Increasing age also made patients more likely to be in this disadvantaged group; whereas higher SES, care in a teaching hospital, and having more documented comorbid conditions decreased the odds. Among the strongest predictors of the quality of care, however, was the patient's level of participation in the healthcare system. Patients who did not regularly see a primary care physician or an oncologist were much more likely to be in the group that did not receive necessary care. Conversely, those survivors who had a primary care physician were less likely, and those who saw both a primary care physician and an oncologist decreased even further their chances of being in the lowest quartile of care receipt. No significant interactions were found in the final model.

Table 5. Multiple Logistic Regression Analysis Predicting Scores Below the 25th Percentile on a Composite Endpoint Measuring Receipt of Necessary Care
VariableaOR95% CI
  • OR: odds ratio; 95% CI: 95% confidence interval; SES: socioeconomic status; PCP: primary care physician.

  • a

    Survivor, compared with controls; Age, for each increasing year of life; African- American, compared with other races; Socioeconomic Status, for each increasing quintile of Socioeconomic Status; Comorbidity, for each point on the Charlson comorbidity scale; Teaching hospital, patients received at least some care in a teaching hospital (compared with none); Neither primary care physician nor oncology specialist, patients saw neither (compared with either); Primary care physician only, patients saw a primary care physician only (compared with seeing other combinations of physician types); Both primary care physician and oncology specialist, patients saw both compared with neither.

African American1.491.31–1.70
Teaching hospital0.830.76–0.91
Neither PCP nor oncology specialist9.317.87–11.05
PCP only0.730.63–0.86
Both PCP and oncology specialist0.370.31–0.43


  1. Top of page
  2. Abstract
  6. Acknowledgements

With only a few exceptions, we found that colorectal carcinoma survivors tend to be less likely than similar controls to receive necessary care across a broad range of chronic medical conditions. There was much less of an effect of survivorship on the management of most acute conditions, as expected. Although it may be argued that a personal history of cancer could affect the appropriateness of some interventions, there is no good rationale for not providing cancer survivors with, for example, appropriate follow-up after an acute myocardial infarction or a stroke. These findings raise the possibility that either a blinding focus on the prior malignancy or nihilism about the prognosis may leave cancer patients' other medical issues relatively ignored.

The Institute of Medicine recently raised concerns about the quality of cancer care in the United States.15 For understandable reasons, most attention has been focused on delivering interventions known to reduce cancer-related mortality and morbidity; however, the noncancer health needs of cancer survivors relatively have been ignored. For most cancer patients, cancer and its treatment constitute only a fraction of their medical history. For the rest of their lives, these patients will contend not only with the risk of cancer recurrence but also with other ailments that may affect their health, longevity, and physical function. The Minnesota Colon Cancer Control Study,16 for example, found that, although fecal occult blood testing decreased colorectal carcinoma mortality by ≈ 33%, there actually was no difference in overall survival because of excess cardiovascular deaths, highlighting the fact that the risk factors for colorectal carcinoma, such as smoking, a high-fat diet, and a sedentary lifestyle, also are risk factors for heart disease. Furthermore, it has been found that noncancer-related comorbid conditions have more influence on overall quality of life for long-term survivors of colorectal carcinoma than their initial stage of disease or time since diagnosis.17 Despite these data, groups such as the American Society of Clinical Oncology have issued surveillance guidelines for several cancers, but they do not mention noncancer-related health concerns.18–20 Even among programs that are devoted to cancer survivorship, noncancer-related health concerns have not been a significant priority.21

Traditionally vulnerable patient populations—the elderly, the poor, and African Americans—consistently experienced poorer quality care on many measures in our study whether they were survivors or not. Our results also suggest that it may be appropriate to add cancer survivors to this list. It is unknown whether the observed patterns of care are driven mostly by patient health-seeking behaviors or by physician diligence. A possible explanation for our observations is that patients with a diagnosis of cancer may lose contact with noncancer providers who are important in their care. Other authors have demonstrated that lack of involvement of a primary care physician decreases the likelihood of receiving appropriate health maintenance.22 The finding that the survivors in our study who were followed by both primary care physicians and oncology specialists were more likely to receive necessary care argues for the importance of keeping these patients broadly engaged in the health care system to be able to maximize their survival and quality of life. It is possible that specialist physicians may not always be aware when their patients expect them to fulfill a primary care role. In fact, a large survey of oncologists found that they generally do not want to take on that role.23 Similarly, primary care providers may assume incorrectly either that there is still an oncologist involved when there may not be or that the oncologist will assume responsibility for all cancer screening, not just for surveillance of the original cancer.

This study, which is the first to look at a broad collection of quality-of-care indicators for a group of cancer survivors, had several strengths. It was population-based; and, as such, we were able to reflect real-world practice among relatively unselected patients. Because all patients were drawn from the Medicare rolls, they all had uniform insurance, removing reimbursement disparities as a potential confounder of the findings. The survivor and control groups were well matched. If anything, the few differences (i.e., living in an urban setting; care in a teaching hospital; and, arguably, the pattern of comorbidity) all were factors that should have favored survivors receiving more health services.3 Moreover, the overall rates for the indicators listed in Table 2 were quite similar to those reported previously in the general Medicare population.6 Although the differences found were small in absolute terms, from an aggregate public health perspective, the total impact on missed clinical opportunities is important. Some may question the utility of some of the screening measures in this very elderly population; however, the evaluations still are consistent with current practice guidelines and demonstrate significant disparities between cancer survivors and those without a history of cancer.

The main limitations of this study were those common to observational studies in administrative data sets. The analyses relied on the accuracy of billing data that were not collected for health services research and, thus, lacked clinical detail and information about patient preferences. Billing practices may not be uniform. Sociodemographic characteristics were known only at the level of the census tract. These results in elderly patients with advanced solid tumors in a fee-for-service environment may not translate to other populations or settings, and survivors age ≤ 65 years and patients who were excluded due to HMO enrollment may have had different experiences of care than what we observed. Finally, it is possible that some of our survivors had recurrent cancer that we were not able to detect. We would argue, however, that their proven indolent disease course still should make them candidates for appropriate management of their comorbid illnesses.

Whether it is due to patient factors, physician factors, or both, cancer survivors appear to be a vulnerable patient population. Having a prior cancer diagnosis may shift attention away from important noncancer problems. In addition, cancer survivors may use specialists as their personal physicians, yet these providers may not always be aware that they are expected to provide increasingly complex primary care. It is important for us to clarify with cancer survivors the relative roles that will be played by primary care physicians and specialists, to remain aware of recommendations for both cancer surveillance and routine preventive health, and to acknowledge any potential nihilistic biases we may harbor to ensure that cancer survivors receive necessary, high-quality medical care.


  1. Top of page
  2. Abstract
  6. Acknowledgements

The authors thank Dr. Steven Asch and his colleagues at RAND for providing the technical documentation for assessing under use of necessary care.


  1. Top of page
  2. Abstract
  6. Acknowledgements
  • 1
    SEER. Persons living with major cancers in the United States, 1998. J Natl Cancer Inst. 1998; 90: 565.
  • 2
    ReisLAG, KosaryCL, HankeyBF, MillerBA, editors. SEER cancer statistics review, 1973–1996. Bethesda: National Cancer Institute, 1999.
  • 3
    Earle CC, Burstein HJ, Winer EP, Weeks JC. Quality of non-breast cancer health maintenance among elderly breast cancer survivors. J Clin Oncol. 2003; 21: 14471451.
  • 4
    Hafner-Eaton C. Physician utilization disparities between the uninsured and insured. JAMA. 1993; 269: 787792.
  • 5
    Kahn KL, Pearson ML, Harrison ER, et al. Health care for black and poor hospitalized Medicare patients. JAMA. 1994; 271: 11691174.
  • 6
    Asch SM, Sloss EM, Hogan C, Brook RH, Kravitz RL. Measuring underuse of necessary care among elderly Medicare beneficiaries using inpatient and outpatient claims. JAMA. 2000; 284: 23252333.
  • 7
    Zippin C, Lum D, Hankey BF. Completeness of hospital cancer case reporting from the SEER Program of the National Cancer Institute. Cancer. 1995; 76: 23432350.
  • 8
    Nattinger AB, McAuliffe TL, Schapira MM. Generalizability of the Surveillance, Epidemiology, and End Results registry population: factors relevant to epidemiologic and health care research. J Clin Epidemiol. 1997; 50: 939945.
  • 9
    Ries LAG, Kosary CL, Hankey BF, Miller BA, Harras A, Edwards BK. SEER cancer statistics review, 1973–1994, National Cancer Institute. NIH publication 97-2789 1997. Bethesda: National Institutes of Health, 1997.
  • 10
    Potosky AL, Riley GF, Lubitz JD, Mentnech RM, Kessler LG. Potential for cancer related health services research using a linked Medicare-tumor registry database. Med Care. 1993; 31: 732748.
  • 11
    Krieger N. Overcoming the absence of socioeconomic data in medical records: validation and application of a census-based methodology. Am J Public Health. 1992; 20: 703710.
  • 12
    Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992; 45: 613619.
  • 13
    Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis. 1987; 40: 373383.
  • 14
    Earle CC, Tsai JS, Gelber RD, Weinstein MC, Neumann PJ, Weeks JC. Effectiveness of palliative chemotherapy for advanced lung cancer: instrumental variable and propensity analysis. J Clin Oncol. 2001; 19: 10641070.
  • 15
    National Cancer Policy Board. Ensuring quality cancer care. Washington, DC: National Academy Press, 1999.
  • 16
    Mandel JS, Bond JH, Church TR, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. N Engl J Med. 1993; 328: 13651371.
  • 17
    Ramsey SD, Berry K, Moinpour C, Giedzinska A, Andersen MR. Quality of life in long term survivors of colorectal cancer. Am J Gastroenterol. 2002; 97: 12281234.
    Direct Link:
  • 18
    Figueredo A, Rumble RB, Maroun J, et al. Follow-up of patients with “curatively resected” colorectal cancer: a practice guideline. BMC Cancer. 2003; 3: 26.
  • 19
    Desch CE, Benson AB, Smith TJ, et al. Recommended colorectal cancer surveillance guidelines by the American Society of Clinical Oncology. J Clin Oncol. 1999; 17: 13121321.
  • 20
    American Society of Clinical Oncology. 1997 Update of recommendations for the use of tumor markers in breast and colorectal cancer. J Clin Oncol. 1998; 16: 793795.
  • 21
    Meadows AT, Varricchio C, Crosson K. Research issues in cancer survivorship: report of a workshop sponsored by the Office of Cancer Survivorship, National Cancer Institute. Cancer Epidemiol Biomarkers Prev. 1998; 7: 11451151.
  • 22
    Rosenblatt RA, Hart LG, Baldwin LM, Chan L, Schneeweiss R. The generalist role of specialty physicians: is there a hidden system of primary care? JAMA. 1998; 279: 13641370.
  • 23
    American Society of Clinical Oncology. Status of the medical oncology workforce. J Clin Oncol. 1996; 14: 26122621.