www.lesswrong.com/posts/sjeqDKhDHgu3sxrSq/sacred-values-of-future-ais

The post claims that in randomized trials where people were assigned “more or less medicine,” those who received more were not healthier on the margin.

However, major randomized experiments on more vs. less medical care (especially the RAND Health Insurance Experiment, which randomized people into plans with different cost-sharing) found some health improvements in the more-generous/free-care groups—particularly for specific conditions and subgroups.

Contradictory evidence from randomized controlled trials

1. NEJM randomized trial (RAND HIE main outcomes paper)

In the NEJM paper reporting results from a randomized controlled trial of free care vs. cost-sharing plans, the authors state that:

• For persons with poor vision and for low-income persons with high blood pressure, free care brought an improvement, including:
  • better vision (reported as 0.2 Snellen lines), and
  • lower diastolic blood pressure (reported as 3 mm Hg)

This is direct evidence that, at least for some participants, the group receiving more medical care (via free care) was healthier on those margins.

2. JAMA randomized trial analysis of hypertension (RAND HIE)

A JAMA paper analyzing hypertension outcomes in the same randomized experiment reports that blood pressures for clinically defined hypertensives were significantly lower with free care than with cost-sharing plans, and attributes this to additional physician contact and improved detection/treatment.

3. American Journal of Public Health analysis of vision (RAND HIE)

An AJPH paper explicitly studied “the improvement in the functional vision of enrollees receiving free care” and concludes that free care resulted in improved vision, via increased eye examinations and lens purchases.

Why this makes the post’s statement incorrect

The post’s sentence is phrased as a blanket claim about the direction of health effects in “lots of randomized trials” (“the people who got more medicine were not healthier on the margin”). But these randomized controlled trial results show health improvements for some outcomes and groups in the higher-care/free-care arms.

It may still be defensible to argue that average effects were small on many metrics (and the NEJM abstract itself notes limited average effects on several measures), but the post’s categorical “not healthier on the margin” claim does not hold given these documented improvements in randomized comparisons.