The Hippocratic oath is one of the oldest codes of professional ethics, and one of its central principles, non-maleficence, is typically expressed as primum non nocere, first do no harm. The idea is altogether good and sensible: most treatments involve some level of risk, so the lowest-risk action—which is sometimes no treatment at all—may be the best course to prevent further harm from befalling a patient. But non-maleficence carries a cost that often goes unexamined. Normally the cost is low and we needn’t worry, but costs can change, and the cost may be very high indeed in certain circumstances. Like, say, a pandemic.
Let me explain with a seemingly unrelated example. Record-breaking wildfires now engulf California nearly every dry season, even though we’ve known for decades that controlled burns would prevent this from happening. Non-maleficence is at work here. Instead of letting low-risk fires burn through the buildup of dry fuel, thereby lowering the risk of cataclysmic wildfires, Cal Fire (which spends $2.1 billion every year on wildfire response) and the U.S. Forest Service (over $1.2 billion every year in California) promptly extinguish every spark, and the wildland tinder continues to grow. ProPublica summarizes it thus:
We dug ourselves into a deep, dangerous fuel imbalance due to one simple fact. We live in a Mediterranean climate that’s designed to burn, and we’ve prevented it from burning anywhere close to enough for well over a hundred years. Now climate change has made it hotter and drier than ever before, and the fire we’ve been forestalling is going to happen, fast, whether we plan for it or not.
But blaming Cal Fire or USFS won’t help. They’re simply responding to their incentives. A low-risk fire ain’t a no-risk fire, and if lives or livelihoods are lost, who do you think the public will blame? When a controlled burn goes awry, we can pinpoint the human error; when yet another megafire breaks out, we’re all too ready to call it an act of god or to blame impersonal systems. We spread the blame thin and no one’s really culpable.
Putting out every low-risk fire and refusing to pursue more aggressive controlled burns demonstrates the flaw in primum non nocere. Non-maleficence on one scale may be quite maleficent on another. What is a perfectly reasonable heuristic for an individual doctor trying to decide upon a course of treatment may in fact be a very bad principle for the institution of medicine.
Some risks are multiplicative, a fact I haven’t seen many health officials or bioethicists publicly acknowledge. Assuming modest (independent) risks early—like recommending masks before we prove their efficacy, or pursuing challenge trials—can keep the multiplicative risk of a deadly, contagious disease from compounding. Outsiders like Alex Tabarrok and Nassim Nicholas Taleb understood the ethical connection. Why didn’t health officials?
Of course unproven treatments are bad. But bad compared to what? The ethical calculus changes when 2,500 deaths a day are heaved onto the scale. When the danger of a disease is multiplicative and the danger of an unproven treatment is not, primum non nocere can be monstrous.
2020 may be remembered as a miracle year for medicine. For what feels like the first time in a long while, we did something fast. But how many more lives could have been saved had we moved even swifter? For six weeks in early spring, the FDA prevented outside laboratories and manufacturers from developing and deploying coronavirus tests. The first batches of the Moderna vaccine were shipped all the way back in February. Between Pfizer’s 22 November application for Emergency Use Authorization and the FDA’s 10 December review meeting, the virus killed another 36,164 people across the country.
Deference, delay, non-treatment—these are still choices, and in a pandemic they take a deadly toll. The risk of additional deaths from an unproven treatment is meaningless unless it’s compared against the human cost of the status quo.