Opinion: Science and the public are learning the wrong lessons from Covid
Get the Health newsletter
Daily health & science — research, biotech, public health, the studies worth knowing. Free.
- Marc Lipsitch and Sara Cody argue in a Los Angeles Times opinion piece that two studies — a book by Princeton's Stephen Macedo and Frances Lee, and a Lancet paper by Tom Bollyky with the IHME at University of Washington — contained methodological errors that biased findings against detecting benefits of COVID-era nonpharmaceutical interventions (NPIs).
- The two studies found little or no evidence that NPIs (masks, social distancing, school and business closures) reduced COVID deaths during the first pre-vaccine year of the pandemic.
- Numerous studies at the time showed NPIs slowed transmission and saved roughly one life for every 150 infections prevented during the early pandemic, according to the opinion authors.
- The opinion authors say the studies' flaw was asking the wrong question — comparing states by date of first stay-at-home order rather than estimating deaths that would have occurred without any intervention; California's order came March 19, 2020, with South Carolina following 19 days later.
- All states intervened within a three-week window, and the states hit hardest early (Northeast urban centers) acted first — biasing the analysis against those early actors because they were 'charged' for infections already spreading when they shut down.
- Co-author Sara Cody, Santa Clara County's health officer during the country's earliest local transmission and first U.S. COVID death, witnessed the benefit of early intervention firsthand and says a proper county-level analysis must account for actual behavior, since adherence to mandates was strongly predicted by politics and trust.
Why it matters: If these flawed analyses become the accepted historical record, the next pandemic will hit the U.S. 'not only unprepared, but saddled with a misunderstanding of what worked and what options may save lives,' the authors warn. The core bias is structural: states intervened within 19 days of each other but at radically different outbreak stages, making 50-state comparisons statistically underpowered to detect any NPI effect — and flipping the conclusion if framed correctly.
