Hollings picked fewer cancer trials — and twice as many hit enrollment targets

A cancer center can only run a fraction of the clinical trials proposed to it, and the choice of which ones determines both patient access and whether its infrastructure is well used. At MUSC's Hollings Cancer Center, the selection process relied on multiple stakeholders and little systematic analysis; high-potential trials were being overlooked while resources went to trials that could not enroll.

Working with operations professor Sanjay Ahire of the University of South Carolina's Darla Moore School, Hollings' clinical and administrative leadership built a multiple-criteria prioritization tool. Physicians, nurses, research teams, and administrators jointly defined more than 15 criteria — clinical importance, patient recruitment potential, resource usage — set their weights, and built rating scales customized to the center.

The result cut against intuition: after implementation, the share of proposed trials selected fell from 26% to 17%, yet the share of selected trials meeting or exceeding patient accrual targets nearly doubled, from 20% to 39%. Selecting fewer, better-scored trials gave more patients access to the trials most likely to matter.

Score every trial on 15+ weighted criteria

The old process let advocacy, not evidence, decide which trials got scarce infrastructure.

Criteria and weights were co-created by the people who run trials, so the scorecard was accepted rather than imposed.

Selecting fewer trials concentrated resources, which is why accrual success nearly doubled.

The tool measured both scientific value and practical feasibility, the two axes advocacy usually trades off badly.

When experts pick projects by persuasion, the fix is not better experts but one shared scorecard built by everyone who has to live with the choice.

Hollings integrated the framework into its standing trial prioritization and selection process. Center leadership credited it with measurably improving patient access to high-impact trials; the work was published in the INFORMS Journal on Applied Analytics in 2026.