Back

Indirect costs and scientific impact at NIMH

Perlis, R. H.

2024-09-24 psychiatry and clinical psychology
10.1101/2024.09.23.24314209 medRxiv
Show abstract

The National Institutes of Health (NIH) awards additional funds for extramural research to support research infrastructure and administration, such that the total cost of a given research project depends on where it is conducted. We sought to understand whether greater indirects were associated with a greater scientific impact of NIH-funded work. The NIH RePORTER database was queried to retrieve all R01, R21, or R03-funded research proposals for which NIMH was listed as the primary funding source for proposals funded between 2012 and 2023. We applied multiple regression to examine the association between indirect rate and measures of scientific impact, including number of publications, their citation impact in terms of H-index per grant and total citations, and the number of patents associated with each grant. Of 5,143 projects, reflecting $9.85 billion, mean indirect rate was 47.9% (SD 16.2%). Greater indirect rate was associated with modest but statistically significantly greater number of publications (+0.30 per 10% increase in indirect rate, 95% CI 0.08-0.51); H-index at 5 years (+0.25 per 10% increase in indirect rate, 95% CI 0.18-0.33; Figure 1); and total citations (+29.71 per 10% increase in indirect rate, 95% CI 17.86-41.57). Each 10% increase in indirect rate was associated with a 20% increase in odds of patent filing (aOR 1.20, 95% CI 1.05-1.37). The results suggest small incremental benefits from conduct of research at higher-cost institutions and provide data for policymakers to consider in weighing the costs against potential benefits of work at such institutions. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=200 SRC="FIGDIR/small/24314209v1_fig1.gif" ALT="Figure 1"> View larger version (20K): org.highwire.dtl.DTLVardef@1abd46borg.highwire.dtl.DTLVardef@6e7b7org.highwire.dtl.DTLVardef@156bb16org.highwire.dtl.DTLVardef@114e2aa_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOFigure 1.C_FLOATNO Multiple linear regression model of association between indirect rate and H-index at five years C_FIG

Published in PNAS Nexus · training set

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.