While scanning the Internet my eye caught an article in GovernmentHealthIT titled The un-health record by Nancy Ferris. It discusses a growing trend by Governments to use health claims data instead of clinical data for a "claims-based EHR". This trend is documented in a report by the US Department of Health and Human Services Office of the Inspector General titled State Medicaid Agencies initiatives on HIT and HIE. Similar initiatives exist in other countries, including Canada, where the Ontario provincial government gives emergency department access to drug claims data for the Ontario Drug Benefit Program.
Its understandable that Governments, with their massive stores of health claims data, would want to put that information to use. However, there is always a risk of using information collected for one purpose (claims adjudication and payment) for another (clinical decision making). Data quality is the issue here.
How good is claims data? From the article:
A 2004 study published in the journal Medical Care found that claim forms showed the correct primary diagnosis slightly more than half the time. For secondary diagnoses, doctor’s offices submitted correct information just 27 percent of the time. Other researchers have come up with comparable findings.
What’s more, claims data lacks some important details and nuance because of the universal coding scheme and the way it is used. For example, the scheme does not distinguish between a severe case of diabetes and one that’s under control, and providers don’t always use the diagnostic codes that indicate the spread of cancers. Furthermore, symptoms such as pain or fever usually don’t show up at all.
So long as health care professionals are fully informed about the limitations of the data, the use of claims data probably brings more benefits than risks. Claims data can be used as one input into the clinical decision-making process. However, in the absence of structured processes for evaluating the quality of the data, and safety risks in eHealth, claims data alone cannot be used as the basis for clinical decision-making.