Big Bad Data, or the case against Electronic Health Records

  The proliferation of available data is generating much excitement in the public health community. However,  this enthusiasm must be tempered by recognition of the potential limitations of EHR data. Source: Big Bad Data: Law, Public Health, and Biomedical Databases EHRs often contain data entry errors, in part because they can increase physicians’ documentation burden. Busy clinicians sometimes type quickly and invert numbers, place information in the wrong patient’s record, click on incorrect menu items, or copy and paste narrative from prior visits without carefully editing and updating it. Much of the information in EHRs is coded using not only the International Classification of Diseases but also customized lists incorporated into EHR products, and coding can introduce further errors. Codes may be confusing, misleading or too general  to indicate the specifics of patients’ conditions. Furthermore, EHRs may not accommodate detailed and nuanced natural language notes about patients’ medical histories and diagnostic findings. Commentators have noted that providers collect data for clinical and billing purposes rather than for public health reasons. Thus, EHR content is not always wellsuited for public health uses. Furthermore, clinicians may have incentives to “upcode” in order to maximize charges, and this practice can systematically compromise the accuracy of many records. The menus and lists built into EHR systems may facilitate upcoding by suggesting items for which physicians should bill and making it easy to click boxes for charge purposes. In some instances, EHRs are incomplete, lacking essential information such as treatment outcomes. Patients who receive medication from their doctors often do not report whether the therapy was effective. The absence of return visits may mean that the patients were cured, but it could also indicate that they failed to improve or deteriorated and decided to visit different doctors or specialists. In addition, patient records are often fragmented. A patient may see multiple doctors in different facilities, and if these practices do not have interoperable EHR systems, pieces of the individual’s record will be scattered in different locations. Such fragmentation can hinder surveillance and research efforts because the patient’s medical history cannot easily be put together into a comprehensive whole. EHR vendors are making slow progress towards achieving interoperability, the ability of two or more systems to exchange information and to operate in a coordinated fashion. In 2010 only 19% of hospitals exchanged patient data with providers outside their own system. Vendors may have little incentive to produce interoperable systems because interoperability might make it harder to market products as distinctive and easier for clinicians to switch to different EHR products if they are dissatisfied with the ones they purchased. The lack of interoperability in EHR systems can also impede data harmonization. Different systems may use different terminology to mean the same thing or the same terminology to mean different things. For example, the abbreviation “MS” can mean “mitral stenosis,” “multiple sclerosis,” morphine sulfate,” or “magnesium sulfate.”15 If the term’s meaning is not clear from the context, then analysts may not be able to interpret it correctly.