The Ledger That Doesn't Balance
The most popular sentence in American health policy, that prevention pays for itself, fails as a general claim, according to the health economists who measure it. Fewer than one preventive measure in five reduces spending, a smoke-free population costs more over a lifetime, and Medicare care-coordination programs showed no savings once fees were counted. The episode keeps what survives: cost-effectiveness, and promises we will keep anyway, bought where cheapest.
Who's who
- Joshua T. Cohen+1 more episodeCo-author of the 2008 NEJM perspective on whether prevention saves money
- Peter J. NeumannCo-author who built the cost-effectiveness registry behind the 2008 finding
- Milton C. WeinsteinCo-author of the 2008 NEJM perspective
- Jan J. BarendregtLead author of the 1997 Dutch study on smoking's lifetime costs
- Pieter H. M. van BaalLed the 2008 replication finding healthy-living cohorts cost the most
- Amy FinkelsteinLead author of the randomized trial of Camden's hotspotting model
- Paula BravemanCo-author of the 2014 statement on social determinants of health
- Chris PopeManhattan Institute critic of the social determinants framing
Where it happens
- Camden, New Jersey+2 more episodesCity whose hotspotting program was randomized in a 2020 trial
- NetherlandsSource of the 1997 and 2008 lifetime cost studies
- United States+16 more episodesSpent 2.9 percent of health spending on preventive care in 2018
Timeline
- September 1994A meta-analysis finds D.A.R.E.'s weighted mean effect on drug use at 0.06.
- October 9, 1997NEJM publishes the Dutch finding that a nonsmoking population would cost more over a lifetime.
- February 14, 2008Cohen, Neumann, and Weinstein report that fewer than 20 percent of preventive interventions reduced spending.
- January 18, 2012CBO reviews 34 Medicare demonstration programs and finds spending unchanged or higher in nearly all.
- January 9, 2020A trial of Camden's hotspotting model finds no reduction in 180-day readmissions.
Still debated
The death-offset result is real but model-dependent, shifting with the list of smoking-related diseases and cost structure. The Camden trial tested one program in one city, so its null is not a null on quality of life, and the correct share of health attributable to clinical care is unknown.