Surveillance Bias: A Threat To Public Quality Reporting and Hospital Acquired Conditions

The More You Look, The More You Find

While everyone agrees that quality monitoring systems should be routinely embedded in health care settings, Elliott Haut and Peter Pronovost, writing in the June 15 issue of JAMA, correctly remind us about the risk of falling prey to "surveillance bias."

The short definition is "the more you look, the more you'll find."  In other words, the greater the number of diagnostic tests, the greater the detection of disease.  Deep venous thrombosis (DVT) is used by Drs. Haut and Pronovost as an example.  Not all patients with DVT have the typical symptoms of a swollen painful leg and not all patients with a positive test have DVT.  Yet, if a hospital decided to initiate a DVT screening program on all its bedbound inpatients, it could end up looking like it had "more" DVT cases than another otherwise identical hospital without such a screening program.

It can get even more complicated.  Noting that DVT is one of CMS' dreaded "hospital acquired conditions," physicians may worry that a surveillance bias could translate into public disparagement or financial penalty.  This could prompt them to under-test for the condition, paradoxically leading to increased patient harm.

The authors conclude with some common sense suggestions:

1) If quality surveillance must be mandated, ensure that the surveillance methods are clearly defined and uniformly applied.  By the way, policymakers will also need to consider the burden of more cost thanks to additional testing.

2)  The risks of all downstream unintended consequences need to be carefully considered, including the perverse incentives that may prompt hospitals to not look and find less.

3) Other measurement methodologies need to be considered.  For example, the authors point out that it may be useful to rely on "process" measures that look back at the care of patients with adverse events.  For example, chart audits of known DVT patients could detect whether any underlying risk factors could have been better managed.

Coda:  Soon after posting this, the DMCB came across this Kaiser Health News report deploring the increase in wrong side surgeries.  Like all modern reportage, KHN has sprinkled in some lurid anecdotes. 

Not to be outdone, the DMCB recalls one surgical colleague's witnessing of an emergency room evacuation of a subdural hematoma (a traumatic blood clot overlying the surface of the brain), which is accomplished by drilling a small hole in the skull.  After the surgeon made the preliminary incision through the skin of the scalp, he looked at the CAT scan one more time and realized it was (like a surprising number of television show x-rays) mounted backwards.  The cut and drilling on the other side was accomplished, but the preliminary incision was counted as wrong side surgery. 

Is the apparent increase in national rates of wrong side surgery partially explained by surveillance bias?  You be the judge.