There is a well done study out in Medical Care by Rebecca Roberts et al. from Stroger/Cook County that assesses the attributable costs of HAIs in a cohort of patients from 2000. What sets this paper apart is the careful attention they paid to design and analysis methodology. Just to paste from the abstract, they used "ordinary least squares linear regression and median quantile regression, Winsorizing, propensity score case matching, attributable LOS multiplied by mean daily cost, semi-log transformation, and generalized linear modeling. Three-state proportional hazards modeling was also used for LOS estimation. Attributable mortality was estimated using logistic regression."
They did all the right things and it's a bit like that Snicker's commercial from years ago - "No matter how you slice it, it comes up peanuts." Costs ranged from $9300 to $21,000 depending on the method used - hey methods matter! Attributable variable costs, which are more likely to be recoverable in the short term, ranged from $1500 to $6800. Excess Length of Stay (LOS) ranged from 5.9 to 9.6 days. Of note, the sample was selected from a higher-risk pool, so their estimates might over estimate the "true" costs.
When their data was extrapolated to estimate national impact of HAIs, they predicted that there are ~350,000 HAIs annually in the US with an impact of $3.2 to $7.3 billion. Not chump change, but far lower than other published estimates. They suggested that 1/3 of infections might be preventable (Mike - What do you think about that?) so that in 2009 dollars, preventing these infections could save $8.4 billion in medical AND societal costs. They also offer a nice comparison amount, since most of us have no idea what a billion dollars represents, when they state that estimated 2008 medical malpractice claims were about $3.6 billion. No word on how many of those claims were for HAIs!
Overall, a great study with great methods. Health economics can be pretty ugly at times, these investigators should be congratulated.
Link: Roberts RR et al. abstract Medical Care November 2010