This news story reports,

Hospitals in Pennsylvania are underreporting a serious infection problem that, by their estimates, last year sickened 11,668, killed 1,793 and added $2 billion to the bill.

… Charles Inlander, a local consumer advocate and member of the Lehigh Valley Business Conference on Health Care, said Pennsylvania hospitals must be underreporting because federal health officials put the percentage of patients who acquire infections in the hospital at 6-10 percent. The 11,668 reported to the state council represent less than 1 percent of the 1.5 million patients admitted in 2004.

… Pennsylvania is the first state to begin collecting and reporting on hospital-acquired infections, also called ”nosocomial” infections.

This data supports Robin Hanson’s hypothesis, also endorsed here by Bryan Caplan, that the reason that health care shows so little benefit on average is that the adverse effects cancel out the positive ones.

Even if Hanson is right, I see no iron law dictating that medical benefit and medical harm must balance. However, it could be that the most cost-effective way to increase longevity in the United States might be to focus on increasing hospital safety.