Can we use biomarkers in combination with self-reports to strengthen the analysis of nutritional epidemiologic studies?

Table 4 Lutein and Age Related Macular Degeneration (ARMD), With Dietary Intake Assessed by 6 24 HR's: Standardized Relative Risks (RR*), Statistical Power and Relative Sample Size** (rss) Required for Various Analysis Strategies Under Different Disease Risk Models

Analysis Strategy		Disease Risk Model
Analysis Strategy		(a) Not mediated through marker	(b) Mediated entirely through marker	(c) Partially mediated through marker
RDI ^a (univariate)	RR Power (rss)	0.43 0.890 (1.00)	0.55 0.629 (1.00)	0.49 0.810 (1.00)
MBL ^b (univariate)	RR Power (rss)	0.47 0.829 (1.20)	0.32 0.995 (0.25)	0.38 0.972 (0.54)
Bivariate^c	RR RDI RR MBL Power (rss)	0.53 0.61 0.904 (0.95)	0.87 0.33 0.986 (0.30)	0.67 0.44 0.968 (0.55)
Howe^d (ranks)	RR Power (rss)	0.38 0.959 (0.74)	0.34 0.985 (0.31)	0.35 0.974 (0.53)
Principal Components^e	RR Power (rss)	0.39 0.952 (0.77)	0.35 0.984 (0.31)	0.37 0.977 (0.51)

* Standardized relative risk is defined as the RR between the 90^th and 10^th percentiles of the distribution.
Bold type indicates the most powerful method among the available choices investigated.
** Sample size requirement relative to the univariate RDI analysis.
^a Regression of ARMD on RDI
^b Regression of ARMD on MBL
^c Regression of ARMD on RDI and MBL
^d Regression of ARMD on combination of RDI and MBL using Howe's method
^e Regression of ARMD on combination of RDI and MBL using Principal Components