Figure 3From: Evidence-based care of older people with suspected cognitive impairment in general practice: protocol for the IRIS cluster randomised trial Potential confounding variables adjusted for in the primary analyses. Back to article page