Table 1. Definitions and Examples of Unstructured Data, Syntactic Variability and Semantic Variability in Discharge Communication.

Category	Examples	Specific Data Elements
Unstructured Data: Generally recorded as free text, with minimal guidelines for content, format or structure of information.	Progress notes Nursing notes Discharge summaries Physical therapy, occupational therapy, and speech language pathology notes History and physical examination Admission assessments Hospital History	Verbs to Describe Warfarin Initiation Utilized, maintained, initiated, started, deployed, observed, begun, etc. INR Assessment Timeframe Weeks, as needed, specific days, per next provider, to be set at skilled nursing facility, absent INR Goal Within therapeutic range, single digits like 2.0, 2.0-3.0, absent How medical follow-up is to be arranged Routine needed, scheduling in process, to be scheduled / determined, defer to another service / provider, communicated with patient, 2 weeks, patient will / is to / should follow-up Medical Follow-Up with Whom and Timeframe PCP but no PCP identified, provider at next facility, notify provider when patient arrives, if PCP feels necessary,rehab unit to arrange, 2 weeks, as needed
Syntactic Variability: Data variability caused by differences in the representation of data elements. These issues are detectable and resolvable using single-site data	Weight, age, sex, birth date may be recorded and stored in different locations, formats or units within an EMR	Multiple written, electronic, and combination of versions with each changing format and location within EMR Service / Provider dependent format for discharge summary Warfarin dose change discussed in narrative text but absent all together on admission/discharge medication list Medical follow-up standard format (date, time, location, and with whom) PCP information automatically generated at the top of discharge summary
Semantic Variability: Data variability caused by differences in the meaning of data elements. Difficult to detect using single-site data alone because data semantics tend to be consistent within an institution	Fasting and random blood glucose, finger-stick or venipuncture, serum or plasma measurements would result in glucose values that do not represent the same concept	Physician signatures Authenticated, hand-written signature, electronically signed, unsigned Multiple locations for discharge summaries Historical encounter, electronic discharge summary link on face sheet, discharge summary tab, notes, scanned PDF files Pro-time versus INR diagnostic test Anticoagulation clinic on site and integrated in EMR process Primary Care Physician Information routinely Located at the top of the discharge summary or manually entered by physician in signature block

PCP = Primary Care Provider; INR = International Normalized Ratio; EMR = Electronic Medical Record