Does a Hospital’s Ranking Increase Patient Volume?

The US News and World Report has popular rankings for colleges, graduate schools and now hospitals.  Do patients actually decide to seek care at different hospitals based on these ratings?  Devin Pope attempts to answer this question in a recent Journal of Health Economics article (free draft).

To determine whether or not this is the case, Pope uses two methods.  First, he employs a fixed effects framework.  This method in essence evaluates how a change in the hospital rank affects the change in the volume of non-emergency patients.  The second method uses a mixed-logit discrete choice model (a.k.a. the random coefficient logit model).  This is based on a random utility framework and is different from the standard conditional-logit model; most importantly, it does not exhibit the logit’s restrictive independence of irrelevant alternatives (IIA) property.

To determine the change in patient volume, the paper focuses on Medicare patients in California.  The reason for this is because Medicare patients have flexible coverage and each hospital receives the same payment from Medicare from each patient.

The paper finds that a hospital-specialty that improves their within state rank by one spot experiences a 6.8% increase in patient volume on average.  The effect is largest for heart problems.  This may imply that patients are most responsive to changes in rankings for heart treatments.

Pope also conducts a number of robustness checks.  One of these is to examine how changes in a hospital’s rankings affect the volume of emergency care.  Since emergency care patients likely have little choice over the hospital to which they are admitted, a change in rankings should have little impact on patient volumes.  This turns out to be the case.


  1. No one bothers if it’s for emergency care I suppose, and if one’s own physician is affiliated with a hospital (s)he may refer you there instead of some more highly ranked option.

Leave a Reply

Your email address will not be published. Required fields are marked *