Predictors of colorectal cancer testing using the California Health Inventory Survey

To identify key variables associated with colon cancer testing using the 2009 California Health Inventory Survey (CHIS).

The CHIS has been conducted biennially since 2001 using a two-stage, geographically stratified random-digit-dial sample design to produce a representative sample of the entire State. For this study we used survey data from 2001-2009 inclusive. We restricted our analysis to White, Black, and Hispanic/Latinos aged 50-80 years. Weighted data was used to calculate the proportion of participants who underwent some form of colon cancer testing (colonoscopy, flexible sigmoidoscopy or fecal occult blood testing) within the previous 5 years stratified by race/ethnicity. For inferential analysis, boot-strapping with replacement was performed on the weighted sample to attain variance estimates at the 95%CI. For mean differences among categories we used t-tests and for comparisons of categorical data we used Pearson's χ(2). Binary logistic regression was used to identify independent variables associated with undergoing some form of testing. Trend analysis was performed to determine rates of testing over the study period stratified by race.

The CHIS database for 2009 had 30 857 unique respondents corresponding to a weighted sample size of 10.6 million Californians. Overall, 63.0% (63.0-63.1) underwent a colon cancer test within the previous 5 years; with 70.5% (70.5%-70.6%) of this subset having undergone colonoscopy. That is 44.5% (44.4%-44.5%) of all individuals 50-80 underwent colonoscopy. By multivariable regression, those tested were more likely to be male (OR = 1.06; 95%CI: 1.06-1.06), Black (OR = 1.30; 95%CI: 1.30-1.31), have a family member with colon cancer (OR = 1.71; 95%CI: 1.70-1.72), and have health insurance (OR = 2.71; 95%CI: 2.70-2.72). Progressive levels above the poverty line were also associated with receiving a test (100%-199%: 1.21; 1.20-1.21), (200%-299%:1.41; 1.40-1.42), (> 300:1.69; 1.68-1.70). The strongest variable was physician recommendation (OR = 3.90; 95%CI: 3.88-3.91). For the Hispanic/Latino group, additional variables associated with testing were success of physician-patient communication (OR = 2.44; 95%CI: 2.40-2.48) and naturalized citizenship status (OR = 1.91; 95%CI: 1.89-1.93). Trend analysis demonstrated increased colon cancer testing for all racial/ethnic subgroups from 2001-2009 although the rate remained considerably lower for the Hispanic/Latino subgroup.

Using CHIS we identified California citizens most likely to undergo colon cancer testing. The strongest variable associated with testing for all groups was physician recommendation.