A total of 207 participants enrolled in the study between July 26, 2006 and November 27, 2007. The number of participants who reviewed an English language prostate or colon cancer brochure or a Spanish language prostate cancer brochure were 17, 62 and 28, respectively. The number of participants who reviewed an English language prostate or colon cancer decision aid or a Spanish language prostate decision aid were 21, 62 and 17, respectively. As shown in there were no significant differences between the brochure and decision aid groups for any of the demographic variables. Also shown in are the ranges for these demographic variables across the 13 participating primary practices. There was significant variability across the clinics making most of the comparisons statistically significant.
Demographics by group and ranges across clinics
shows participants’ role preferences by group. The results suggest that participants who reviewed a video decision aid were more likely to desire being the primary or sole decision maker, with three of five questions showing statistically significant differences between the two groups, even after applying a Bonferroni adjustment for multiple comparisons. We explored whether there were any differences by language among participants who reviewed a prostate cancer brochure or decision aid, but did not find any significant effects (data not reported).
Our knowledge hypothesis was confirmed. As shown in participants who reviewed a decision aid had higher knowledge about prostate and colon cancer. The effects were significant using analysis of variance (Prostate cancer: F(1,82) = 9.36, p=.003; Colon cancer: F(1,120) = 11.22, p=.001) and using linear mixed models controlling for clustering in clinics (Prostate cancer: F(1, 79.73) = 11.23, p=.001; Colon cancer: F(1, 115.99) = 11.15, p=.001). The intra-class correlations were .03 and .09, for prostate and colon cancer knowledge, respectively. There were no significant differences by language among participants who reviewed a prostate cancer brochure or decision aid.
Participants’ intentions to work with their doctor to make a cancer screening decision were significantly correlated with attitudes (ρ = .37, p<.001), perceived social norms (ρ = .51, p<.001) and self-efficacy (ρ = .53, p<.001). Together, these constructs accounted for 30.5% of the variance in behavioral intention (F(3,154)=22.07, p<.001, R2=.305). shows the Integrative Model measures comparing brochure and decision aid groups, broken down by prostate and colon cancer screening. The table shows the statistics for univariate and multivariate main-effects and interaction effects, as well as the intra-class correlation, which indicates the degree of clustering by clinic. All models controlled for participants’ language. There were no significant differences in attitudes comparing the four groups. Contrary to our hypotheses, there were significant univariate and multivariate main effects indicating that participants who reviewed a brochure had stronger perceived social norms, greater self-efficacy and greater behavioral intentions about working with their physician to make a cancer screening decision. In addition, there were significant univariate interaction effects for each of these measures, suggesting that the prostate cancer decision aid produced the lowest scores. The multivariate interaction effect for perceived social norms was also significant, however, the multivariate interaction effects for self-efficacy and behavioral intentions were only marginally significant. The degree of clustering by clinic for these measures was limited. shows the Integrative Model measures by language for participants who reviewed a prostate cancer brochure or decision aid. The main effects of brochure versus decision aid remained, but there were no significant effects for language or interaction effects.
Integrative Model measures by language - Prostate cancer only
In the brochure group 57.1% of participants reported talking to their doctor about cancer screening compared to 45.8% in the decision aid group. The comparison was non-significant (χ2(1)=1.21, p=.272). In the brochure group 49.5% of participants reported choosing cancer screening, 41.7% decided not to screen and 8.7% reported still feeling unsure after seeing the physician. In the decision aid group 38.8% reported choosing cancer screening, 49.0% decided not to screen and 12.2% reported feeling unsure. The differences between the two groups were non-significant (χ2(2)=2.48, p=.289). shows the proportion of participants who reported talking to their physician about cancer screening by their screening decisions. There were significant differences (χ2(2)=27.53, p=.000), which remained after controlling for clustering (t(8)=3.49, p=.008), in how many participants spoke with their physicians depending on their screening decisions. There was no effect of language and no interaction effect. Participants who chose not to screen were least likely to talk to their physician about screening. There were also significant differences in participants’ intention to work with their physician to make a decision comparing different screening decisions (F(2,194)=6.58, p=.002). There was no effect of language and no interaction effect. These effects remained in multivariate models that controlled for patient clustering in clinics (F(2,180.82)=7.78, p=.001). Participants who chose not to screen (M=1.79, SD=1.69) or remained unsure (M=1.84, SD=1.28) had lower intentions to work with their physician to make a decision prior to entering the consultation than participants who chose screening (M=2.56, SD=0.89).
Proportion of participants reporting talking to physician by screening decisions