Ranked Differences between Expected and Actual NIH Funding According to Year of Funding and United States Disease Burden Measure(s) Used.

*Ascending rank, from most underfunded disease condition (indicated by negative numbers) to most overfunded (positive numbers) as predicted by each model.† Univariate linear regression of the association between 2004 disease-specific Disability-Adjusted Life-Years (DALYs) and the outcome NIH dollars. Differences between expected and actual funding levels in 1996 are adjusted for inflation to 2006 dollar equivalents, but are otherwise unchanged from those reported by Gross et al.‡ Standard univariate linear regression of the outcome NIH dollars as predicted by disease-specific DALYs. A stepwise forward multivariable model retained only DALYs as a predictor.§ Standard multivariable linear regression of the outcome NIH dollars as predicted by disease-specific DALYs and charity revenue.¶ The constrained model is the multivariable linear regression model where the predicted NIH dollars are obligated to be proportional to disease-specific DALYs after adjustment for total number of hospital discharges and average hospital charges.