Skip to Content
National Cancer Institute U.S. National Institutes of Health
Colorectal Cancer Mortality Projections: Modeling the impact of interventions on US cancer mortality

Risk Factors for Colorectal Cancer

Sections on this page:

Summary: learn which risk factors we modeled and how

Reducing risk factors for developing colorectal cancer (CRC) is a critical part of any comprehensive cancer control effort. However, results are hard to measure because of the number of risk factors involved, the variable quality of the data available to track risk factor trends, and the long latency period between exposure to a risk factor and development of the disease.

Some risk behaviors (smoking, primarily) have been improving over the recent past, while others (obesity and exercise) have been getting worse. Model results show that overall, if current trends continue (including positive trends in screening and chemotherapy), CRC mortality will continue to decline. By making an additional effort to reverse unfavorable risk factor trends and increase positive ones, we gain a small additional reduction of projected CRC mortality. There are several reasons:

  • The goals we set for additional change are small, because risky behavior is difficult to change.
  • The long period between the onset of an adenoma and its possible development into cancer (10 to 30 years, depending on the model) means that changes in risk factor prevalence take a long time to show up in CRC mortality rates. The change in mortality may be more dramatic after 2020.

  • This simulation measures only the effect of risk factors on CRC. We did not account for decreases in mortality from other types of cancer, heart disease, diabetes, and other diseases.

If we could meet all the Healthy People 2010 (HP 2010) risk factor reduction objectives, the change in mortality would be more dramatic. However, most of the HP 2010 objectives seem out of reach. Also, these objectives were intentionally set without regard to historic race and gender differences, so meeting them will be harder for some groups than others.

Risk Factors Included in the Model

The two simulation models incorporate the effects of eight modifiable risk factors known to be associated with colorectal cancer. Those marked with an asterisk (*) have related Healthy People 2010 objectives.

  • Smoking status* (yes/no)
  • Obesity* (based on body mass index (BMI))
  • Physical activity* (met-hours per week)
  • Fruit and vegetable intake* (servings per day)
  • Multivitamin use* (yes/no)
  • Red meat intake (servings per day as a main dish)
  • Aspirin/non-steroidal anti-inflammatory use (yes/no)
  • Postmenopausal hormone replacement therapy (HRT) (yes/no)

Smoking, obesity and red meat consumption increase the chance of developing CRC, while physical activity, fruit and vegetable consumption, multivitamins, aspirin, and HRT have a preventive effect.

We attempted to isolate the effects of specific risk behavior using these models but that is difficult because the effects are so small in the time period studied. Both models indicated that obesity, diet and exercise have a greater impact than smoking or multivitamin use; one model also showed a noticeable positive effect of taking multivitamins. But because the results were too small to be reliable, we only present projections for the combined effect of all eight risk factors.

Return to top

We used four waves of the National Health and Nutrition Examination Surveys (NHANES I, 1971-1975; NHANES II, 1976-1980; NHANES III, 1988-1994; 1999-2002) to estimate changes in CRC risk factors over time. We projected changes into the future under three assumptions, which we labeled Projected Trends, Optimistic but Realistic, and Healthy People 2010 Goals Met. See the section on Scenarios for more detailed information about these assumptions.

Smoking has been declining in the US for some years, and is likely to continue to decline. For all demographic groups except black males, the HP 2010 objective of 12% seems within reach and our Optimistic scenario assumes that a level of 8% is realistic in that time frame. For black males, our optimistic assumption is that smoking will fall to 15%.

Obesity, Diet and Exercise measures have been stable or going in an unfavorable direction in recent years. The obesity problem and the related exercise problem are well known. After falling in the 1970’s and 80’s, red meat consumption is rising slightly. Fruit and vegetable consumption is stable or falling slightly. None of the associated HP 2010 objectives are realistically reachable. In our models, Optimistic but Realistic goals seek an improvement of about 4 percentage points over current levels.

Multivitamin use by white males and females has been increasing steadily, and is currently at about 50%. Use among black men and women has leveled off at around 25% to 30%. HP 2010 sets a target of 80% for women of child-bearing age, for the benefit of pregnancy and childbirth. There is no HP 2010 objective explicitly for multivitamin use for cancer prevention.

Aspirin/non-steroidal anti-inflammatory drug (NSAID) use has been shown to be beneficial in reducing colorectal cancer. However, because of potential side effects and complications from its use, we did not set an Optimistic but Realistic goal for this risk factor, and Healthy People 2010 does not either. We used projected trends for all demographic groups—around 50% for whites and 20% for blacks.

Although postmenopausal hormone replacement therapy (HRT) is associated with a lower risk of colorectal cancer and of bone fractures, it may also be associated with an increased risk of cardiac events such as heart attack, stroke, and blood clots and an increased risk of breast cancer. Trial results suggest that overall, the risks outweigh the benefits [Rossouw et al. 2002]. Accordingly we did not set an Optimistic goal for post-menopausal HRT use. Healthy People 2010 does not have an objective for it either.

For more discussion of current trends for many of these behavioral factors, see the Prevention section of the Cancer Trends Progress Report.

Return to top

Model Inputs for Risk Factors

The following table shows the risk factor model inputs in 2010 for each race/sex combination. Items marked with an asterisk (*) vary from the corresponding Healthy People 2010 objective. For instance, the target for use of multivitamins is derived from an HP 2010 objective targeted to women of childbearing age, to assure that they get enough folic acid. Here we apply it to the general population.

Risk Factor Input Race/Sex Group
(Click for graph of input projected through 2020)
2000 Level Level in 2010
Projected Trends Optimistic but Realistic HP 2010
(Click for more about the objective)

* Statement of objective varies from Healthy People 2010
   
   
** Value in parentheses is the Projected Trend level. An optimistic goal was not used because the risks from aspirin/NSAID use may outweigh the benefits.

% current smokers
white male 24% 15% 8% 12%
white female 20% 15% 8%
black male 35% 22% 15%
black female 21% 16% 8%
% obese
white male 28% 34% 30% 15%
white female 34% 45% 41%
black male 28% 33% 29%
black female 52% 62% 58%
% healthy weight
white male 23% 20% 23% 60%
white female 35% 27% 31%
black male 28% 22% 26%
black female 16% 12% 14%
% moderate or vigorous physical activity
white male 33% 32% 35% 50%
white female 30% 30% 33%
black male 38% 36% 38%
black female 28% 29% 32%
% vigorous activity
white male 24% 24% 26% 30%
white female 21% 20% 23%
black male 33% 30% 33%
black female 21% 21% 26%
% no leisure-time physical activity
white male 36% 37% 32% 20%
white female 36% 38% 33%
black male 40% 46% 41%
black female 56% 56% 51%
% eating 5+ servings of fruits / vegetables per day *
white male 44% 44% 47% 50%
white female 36% 36% 39%
black male 36% 38% 41%
black female 33% 33% 37%
% using multivitamins *
white male 42% 56% 62% 80%
white female 45% 51% 58%
black male 24% 24% 30%
black female 29% 31% 37%
% receiving hormone replacement therapy
white female 18% 9% n/a
(9%)**
n/a
black female 5% 3% n/a
(3%)**
% eating 2+ servings/week of red meat
white male 30% 31% 29% n/a
white female 23% 24% 21%
black male 31% 34% 32%
black female 24% 24% 21%
% Aspirin/NSAID users
white male 42% 50% n/a
(50%)**
n/a
white female 44% 49% n/a
(49%)**
black male 19% 19% n/a
(19%)**
black female 25% 22% n/a
( 22%)**

Return to top

How Risk Factors are Incorporated into the Model

We used the Nurses Health Study (NHS) and the Health Professionals Follow-up Study (HPFS) to estimate the effect of risk factors on colorectal cancer incidence (for more details about the studies, see References and Data Sources). These studies do not show development of adenomas directly, of course, but we estimated the effects of CRC risk factors on unobserved states of colorectal disease by creating simulations of the past using the models and comparing the simulated results with analyses from the NHS or the HPFS. We then ran these simulations in an iterative fashion, revising the models' calculations of risk factor effects, until diagnosed colorectal cancer in the simulated dataset matched the analogous observation in the cohort studies.

The models are designed to replicate the distribution of risk factors across the population. For each simulation run, we input the appropriate assumptions about future risk factor prevalence—projected trends by default, or one of our two scenarios: Optimistic but Realistic, or Healthy People 2010 Goals met (see Model Inputs, above).

We use a risk profile in determining the development of adenomas and CRC. Risks are modified upwards or downwards to reflect the lifestyle risk factors listed above and age, which is itself an important risk factor for CRC.

The relative risks for adenomas and for CRC due to lifestyle risk factors are shown below, for each model. Values less than one signify a decrease in personal risk of CRC, while those over one signify increased risk. In the SimCRC model risk factors can influence both adenoma incidence and progression to stage 1 CRC, whereas in the MISCAN model risk factors only influence adenoma incidence.

Risk Factor for CRC SimCRC MISCAN
Relative Risk for Adenomas Relative Risk for CRC Relative Risk
Smoking (smoker yes/no) 1.60 1.64 1.48
Obesity (BMI 30+) 1.03 1.11 1.33
Physical activity (20+ MET-hrs/week) 0.94 0.87 0.73
High vegetable consumption (5+ servings/day) 1.00 0.99 1.00
Red meat (2+ servings/week as a main dish) 1.03 1.07 1.33
Multivitamin use 0.42 0.54 0.63
Aspirin/NSAID use (~2+ tablets/week) 0.31 0.38 0.63
HRT use (post-menopausal women only) 0.76 0.68 0.73

The two models have other slight differences in the way risk factors are incorporated.

In the SimCRC model:

  • At birth, each simulated person is randomly assigned a vector of risk factor values. Risk values are assigned to reflect observed risk factor distributions by age, sex, race, and calendar year.
  • Each year the risk factor values assigned to a simulated person are updated. For example, BMI is allowed to increase or decrease as a function of age, sex, race, and calendar year to reflect population changes observed in national surveys. Correlations among risk factors are considered (e.g., smokers tend to have lower BMI values compared with non-smokers).
  • The percent of the simulated population with a particular risk factor in each year is determined by the percentage of the population with each risk factor (e.g., BMI of 30 or greater for obese) – age-adjusted to a 2000 standard population.

In the MISCAN model

  • Each simulated birth cohort is assigned a risk factor profile. This risk profile is updated with age to reflect changes in risk factor prevalence by age.
  • The average risk of a cohort to develop adenomas and CRC depends on the risk factor profile; it is the product of the risk factor prevalence and the associated relative risk of these risk factors. The relative risks were based on the published odds ratios found in NHS and HPFS.

For more information about the models, see Simulation Models.

Return to top

 Last modified: December 21, 2010 Accessibility | Privacy Policies | File Formats   
CISNET Home National Cancer Institute Department of Health and Human Services National Institutes of Health USA.gov