Evaluation of Measures and a Brief Intervention to Increase Colorectal Cancer Screening

This study examined i) the reliability of the Health Belief Model (HBM) factors related to colorectal cancer screening among the average-risk, low-income population in non-clinical settings, ii) the invariance of the measures across race/ethnicity and gender, iii) the association of HBM factors with outcomes (colorectal cancer screening) and iv) the effectiveness of small group education in increasing use of colorectal cancer screening. The first step was to determine the appropriateness of the Health Belief Model for use with assessing factors related to colorectal cancer screening. A confirmatory factor analysis (CFA) indicated that the proposed model did not fit the data well. A re-specified model was developed and tested, where several poorly performing items were dropped and residual correlations of 12 items were used. The re-specified model indicated a better fit with the RMSEA = 0.040 (95% CI, 0.036 - 0.044), CFI = 0.965, TLI = 0.958 and SRMR = 0.038. The Chi2(193) was 521.554, p\u3c0.0001 and the Scaling Correction Factor was 1.2386 for MLR. A series of progressively stricter invariance tests were employed, which indicated the presence of weak (metric) invariance across the two racial/ethnic groups (Blacks and Hispanics) indicating that the factor loadings were measured equally by the measures. However, the colorectal cancer screening measures were fully invariant by gender (males and females). Therefore, it was determined that the measures based on the re-specified Health Belief Model (HBM) of colorectal cancer screening discussed in this study provide a reliable and valid way of measuring colorectal cancer screening behaviors of average-risk, low-income individuals in a non-clinical setting. Future studies should focus on examining the measurement invariance across all racial/ethnic groups and gender groups, including the transgender population, to be able to generalize the findings across larger populations. Measurement invariance across different language groups can be another area of future research. A multi-level multivariate regression method was employed to examine the association of HBM factors with the outcome (colorectal cancer screening). The results indicated that when adjusted for race/ethnicity and educational attainment, only the perceived barrier factor was statistically significantly associated with screening; this finding is consistent with previous studies that were conducted among clinical populations. It should be noted, however, that the types of barriers may vary across different studies. Interventions aimed at addressing perceived barriers may be helpful in increasing colorectal cancer screening among low-income, average-risk individuals in non-clinical settings. Future studies may focus on identifying the types of barriers in specified populations that impact colorectal cancer screening rates. Effectiveness of a small group education intervention was examined using multi-level structural equation modeling, which indicated that the intervention was not more effective than the brief instruction, although the program was effective to get more than 55% of participants screened. We speculate that poor socio-economic status, shorter length of intervention and attrition may have played some role. Future studies may consider examining the effectiveness of different lengths of small group education