Language was not associated with colposcopy consent comprehension after adjustment (b=-0.44, p=.46)
Source
Krankl (2011). Patient predictors of colposcopy comprehension of consent among English- and Spanish-speaking women. Womens Health Issues.
Description #
After adjusting for confounding variables (notably education and other socioeconomic factors) in a robust multivariate linear regression, language was NOT significantly associated with colposcopy comprehension (β = −0.44; 95% CI −1.62 to 0.73; p = .46) (Table 3). This is a null result: the large unadjusted English–Spanish gap disappeared once confounders were controlled, and the authors caution that the sign and magnitude of β cannot be interpreted because the change is not significant.
"Despite lower scores on the colposcopy knowledge survey among Spanish-speaking subjects, after controlling for confounding factors in multivariate analysis, language was not significantly associated with diminished comprehension (β = −0.44; p = .46). The value and direction of β cannot be interpreted because the change in summary colposcopy knowledge score is not significant." (Krankl, 2011, p. 83)
"Using linear regression to adjust for confounding variables, we found that language was not significantly associated with greater understanding (p = .46)." (Krankl, 2011, p. 80)
Methods Context #
What? #ⓘ
The observable: the adjusted (independent) effect of language group on the summary colposcopy knowledge score, expressed as a regression coefficient with 95% CI.
"Multivariate model: adjusted effect of language and education … Language −0.44 −1.62 to 0.73 .46" (Krankl, 2011, p. 84)
How? #ⓘ
Robust multivariate linear regression starting from language as the covariate of interest, adding each a priori confounder (retained if p < .05) to estimate language's absolute effect on the outcome.
"Multivariate analysis was used to examine the relationship between subject understanding of colposcopy and multiple potential predictive variables using robust linear regression." (Krankl, 2011, p. 82)
Who? #ⓘ
The 149 complete-data colposcopy patients (111 English, 38 Spanish speakers) at two Boston hospitals; reference category for language was speaking Spanish.
"This study actually accrued 111 English speakers and 38 Spanish speakers." (Krankl, 2011, p. 82)