This article is more than 5 years old. Information may no longer be current.
Response to IRIS Registry article
Click Here to Manage Email Alerts
To the Editor:
The headline for the cover story in the March 10, 2014, issue of Ocular Surgery News is “IRIS Registry to allow wide-ranging access to accurate ophthalmology data.” The IRIS project is a thrilling one. It is not, however, without serious concerns. There is a word in the headline that certainly needs to be challenged, specifically “accurate.”
There is a great deal of evidence to show that not only are paper charts not accurate, but electronic health records are also inaccurate. Indeed, there seems to be more errors on electronic records than on paper records, largely because of the “copy” ability, which means that it is routine to transport information from one examination to the next, whereas something has changed, but it is not captured accurately in the record.
George L. Spaeth
Big data does not necessarily mean accurate data. That is well known to the experts who deal with big data. However, the effect of systematic bias is known to very few except those who are truly expert in the field of statistics and have had the benefit of being instructed by somebody who understands the problem. Big data does not eliminate the problem of systematic bias unless methods to try to mitigate that problem are consciously and successfully integrated in the methodology. For example, whenever we do a procedure or start a medication, we hope the results will be favorable, as does the patient. Therefore, when we look for markers after the medication has been started or the procedure performed, our hopes bias us when we make measurements. Prior to the surgery we may round up and after the surgery we may round down with regard to IOP. Prior to the surgery we may be more responsive to the patients’ concerns about how they feel and after the surgery may be less concerned about their symptoms. In fact, every study that has been done shows that type of bias is absolutely routine. Therefore, the results of big data are no more likely to be accurate, that is valid, than when fewer results are tabulated.
Having a national database is enormously important, and all those who are involved in the creation of the Intelligent Research In Sight (IRIS) Registry should be congratulated. It is a magnificent step ahead. However, it has the potential for misleading as well as the potential for glorifying. It is not the registry itself that will answer that. It is how the registry is operated and the results interpreted.
George L. Spaeth, MD
OSN Glaucoma Board Member
Wills Eye Institute
Jefferson Medical College