Skip to main content

Receipt of CRC Screening/Testing

Each time a person is recommended to take a test or receive treatment, the model must decide whether or not the individual complies with the recommendation. The model uses different methods to make this decision depending on the test and situation.

Receipt of Routine Screening

Never compliant

Some people in the population never undergo any screening test. These people are identified randomly using a probability specified by the user. This probability can be different for each population defined in the model.

Initial compliance

The initial compliance probability is the probability that a person completes a routine screening test at the earliest age he or she is recommended to be screened. The model supports two methods for calculating the initial compliance probability:

  • Direct entry – The user specifies the probability directly. The probability can be different for each population defined in the model. Within a population, all people are assigned the same probability.
  • Statistical model – A multilevel logistic regression model was used to estimate the probability that a person turning 50 years old will comply with routine screening, either FOBT or colonoscopy, at any time point during the 6-year study period. The statistical model used individual characteristics (e.g., sex, race, distance from zip code to nearest endoscopy facility) and geographic characteristics at the zip code level (e.g., population-adjusted number of medical generalists, percentage of residents below federal poverty level) specific to each person.

The estimated 6-year compliance probabilities, p, from the statistical model were converted to either 1-year (for FOBT) or 10-year (for colonoscopy) compliance probabilities within the simulation: p1 = 1 – (1 – p)^(1/6) and p10 = 1 – (1 – p)^(10/6).  These calculated values are mathematically equivalent to those that would be obtained by converting the 6-year probability to an instantaneous rate and converting the rate to 1- or 10-year probabilities [1]. This conversion assumes that the instantaneous rate of compliance is constant over time.

The choice of method can be different for each population defined in the model.

Repeat compliance

The repeat compliance probability is the probability that a person completes a routine screening test at a recommended age other than the earliest age. The model uses the initial compliance probability at every recommended age, regardless of compliance history, unless a polyp has been removed. In the event that a person had a polyp removed, the individual will have a higher rate of compliance since he or she will be aware of the increased risk for CRC and be more likely to adhere to screening guidelines.

Effect of Population Intervention on Probability of Routine Compliance

It is possible to study the impact of population interventions to increase compliance for routine screening using the model. For each intervention studied, the routine compliance probability is increased by a specified fixed amount. The increase in compliance probability may be race-specific or it may be location-specific. The interventions only affect the probability of routine compliance.

Diagnostic Compliance

The diagnostic compliance probability is the probability that a person takes a diagnostic test when recommended to undergo a diagnostic test. The user specifies this probability directly. The probability can be different for each population defined in the model. Within a population, all people are assigned the same probability.

Surveillance Compliance

The surveillance compliance probability is the probability that a person takes a surveillance test when recommended to undergo surveillance. The user specifies this probability directly. The probability can be different for each population defined in the model. Within a population, all people are assigned the same probability.

Treatment Compliance

The treatment compliance probability is the probability that a person receives cancer treatment when clinically diagnosed with cancer. The user specifies this probability directly. The probability can be different for each population defined in the model. Within a population, all people are assigned the same probability.

Sources:

  1. Briggs A, Claxton K, Sculpher M. (2006). Decision modelling for health economic evaluation. New York: Oxford University Press.