Basically, Coding Clinic Q4 2015 received a question if physicians could use a code number in lieu of a diagnosis. Their answer was a resounding no--but they went a little further.
Coding Clinic told the inquirer that it was not appropriate for providers select a code number from a list of codes in place of a written diagnostic statement.
Their rationale is that the ICD-10 is a statistical classification, not a diagnosis. Further, they indicated that many clinical diagnoses may be included in an ICD-10 code, and it may be important to document these diagnoses. They went on to say that it was the provider's responsibility to provide clear documentation of a diagnosis.
The abbreviated ICD-10 descriptions are often so shortened they make no sense--and can't be called a diagnosis. One reason that writing a code in lieu of a diagnosis is inappropriate is that the code chosen may simply be wrong--without a narrative description (of sufficient specificity), there's no way to even know if it's right. Coding is done from a narrative, not from a number.
This issue was very common in ICD-9. Think back to all of the "Diabetes with _____________manifestations". These codes were chosen all the time, and could not be supported--there was no way to know what the "manifestation" was, without further description by the physician. A manifestation isn't a disease description--it's a whole array of possible diseases. Without further documentation by the physician, it's not possible to know what that disease is. In ICD-10, with its added complexity, this issue will arise over and over again. It is the clinician's responsibility to ensure that their documentation makes sense, and includes sufficient information to validate that the code chosen is correct. The act of choosing a code is not enough.
EMRs are great time savers. But they are not a substitute for adequate documentation by the physician to determine if the selected code is correct.