What this covers

Adam Cifu, a physician and author at the University of Chicago, speaks with Russ Roberts on EconTalk about what he calls Medical Conservatism—a clinical approach that resists adopting new therapies until evidence for them is robust, unbiased, and demonstrably improves outcomes that matter to patients. The conversation rests on three interlocking observations: that most medical advances over a 25-year career are small incremental improvements rather than transformative breakthroughs; that industry funding and design of clinical trials introduces systematic bias into the evidence base, making "published" and "true" fundamentally different things; and that clinicians often overestimate their power to affect outcomes, mistaking the body's own healing capacity—and the natural resolution of cyclical conditions—for the fruits of intervention.

Cifu and Roberts explore how these dynamics play out across several domains. They examine the tension between Content Expertise vs Critical Appraisal, arguing that specialist enthusiasm for an intervention often crowds out the statistical rigor needed to judge whether it actually works. They discuss how Population vs Individual Benefit diverge—screening programs and preventive therapies may produce tiny benefits for populations while generating false alarms and surrogate outcomes of dubious relevance for individuals. The conversation touches on the Replication Crisis through the striking case of studies on the SLC6A4 gene and depression, where hundreds of peer-reviewed papers later proved meaningless. Cifu argues that thoughtful patients should slow down and ask what alternatives exist and what "help" concretely means, while physicians should recognize humility about their own effect as protection against hubris. The discussion also acknowledges the role of Placebo Effect and the importance of physician attention as teachable clinical tools often neglected in the rush toward new interventions.

Sharpest takeaway

Cifu argues that being a 'medical conservative'—adopting new therapies only when benefits are clear, evidence is strong and unbiased, and outcomes that matter to patients are improved—is the ideal approach to care, because most new interventions offer only small incremental gains, much published evidence is industry-biased or false, and the human body's own healing capacity means clinicians affect outcomes far less than hubris suggests.

  • Most advances over a 25-year career are small incremental improvements over existing therapies, not transformative breakthroughs
  • Industry funds most trials and is skilled at designing/selling data, so 'published' does not equal 'true'
  • Content experts are enthusiasts for their interventions and conflate experience with evidence, whereas conservatives privilege critical appraisal of proven patient benefit

The argument · threads4 threads · 31 claims
0.85

Distorted patient incentives from low out-of-pocket costs drive overuse of marginally beneficial or wasteful medical interventions.

2 pointscentrality 5/5
  • Small incremental improvements—such as a drug absorbed more easily or taken less often—are awarded patent monopolies and priced many-fold above generics, then preferentially prescribed even though the system fails to encourage patients toward equally effective generics, which is a serious flaw of the current arrangement.

    “a patent is given out, say, because it's absorbed more easily in the stomach... It is many-fold times the cost, say, of a generic.”

  • Because patients pay little out of pocket today versus 50-60 years ago, their incentive to weigh costs against benefits is distorted: small positive benefits relative to a near-zero personal cost lead patients to accept tests and treatments that are wasteful or immoral at the system level even when the social cost is enormous.

    “our incentive as patients to take account of costs and benefits is very distorted. Our costs are often small. The benefits could be tiny, but they are positive relative to a very small cost.”

0.71

Doctors and patients systematically misattribute natural recovery and cyclical improvement to medical interventions rather than time.

2 pointscentrality 4/5
  • Medical conservatives resist conflating a therapy's benefit to a population with its benefit to an individual; many screening tests like prostate screening and mammography deliver shockingly small or zero benefit while generating enormous numbers of false positives that alarm patients and lead to dangerous downstream tests and treatments.

    “We resist the urge to conflate benefits of a therapy to a population vs benefit to the individual.”

  • Because many conditions are cyclical and resolve on their own with the passage of time, both patients and doctors systematically misread which treatments work, attributing natural recovery to an intervention—as when a steroid shot appears to cure a shoulder problem that, like a later untreated episode, would have resolved regardless.

    “a lot of things are cyclical. They get better on their own and we misread what treatment works and doesn't work.”

0.71

Scientific publications and peer review do not reliably establish truth; major bodies of studies frequently prove meaningless or wrong.

2 pointscentrality 4/5
  • Roughly 450 to 1000 studies in the 1990s and beyond confirmed a link between the gene SLC6A4 and depression, but a recent study found no medical basis for the connection, suggesting that a large body of studies that passed all standard tests was meaningless—a major wake-up call about scientific reliability.

    “Maybe 450 or maybe even a thousand studies were done confirming this relationship. A new study has come out finding there is no medical basis for that connection. It's a sham. A false positive.”

  • Publication of a scientific finding does not establish its truth; 'published and true are not synonyms,' because peer-reviewed and even well-designed studies frequently turn out to be wrong over time.

    “"Published and true are not synonyms."”

0.57

Aligning decision-maker incentives with patient welfare through skin in the game beats relying on actors whose interests diverge.

2 pointscentrality 3/5
  • By analogy to Taleb's point that you should not consult the carpenter who built the roulette wheel to understand gambling odds rather than a statistician, deep practical familiarity with an intervention does not confer the statistical understanding needed to judge its true value.

    “when you are trying to understand gambling... you really don't want to use the carpenter who built the roulette wheel on the grounds that, 'Well, the carpenter knows more about it than you do.'”

  • Pharmaceutical companies are not evil but rational actors responding to incentives, so good policy requires giving the relevant decision-makers 'skin in the game' that aligns with patient welfare rather than relying on parties whose incentives work in the opposite direction.

    “drug companies and others are not--are not--evil, malicious folks. They are just responding to the incentives.”