Contact us:
Phone: 734-905-3600
email:
How does your current system measure up?
In hospitals without an EMPI, an average of 18% of records are duplicates.
Even those
with a matching system in place still average 12%.
As hospitals continue to move toward EHRs and work more closely with ACOs, CINs, and OSCs,
the rate of duplicate records increases exponentially.
The system you choose to handle this problem is critical. A difference of just
1% can prevent adverse events and save millions of dollars in costs. What are you using?
- • Basic matching: first/last name + date of birth
- • Maintains a very low false positive rate
- • Poor at handling incomplete or flawed data
- • Misses 26% of duplicates
- • Requires high rate of human intervention
- • Characterized by an exponentially increasing set of complex conditional tests
- • Features an increasingly complex set of conditional statements.
- • Linkage rules are typically implementation specific
- • Misses 8% of duplicates
- • Applies Fellegi-Sunter style algorithm to calculate probability of a match across fields
- • Naïve about dependent probabilities and social constructs, e.g. twins, Jr./Sr., address/phone
- • False positive matches become problematic
- • Misses 5% of duplicates
- • Can be tuned for implementation/regional-specific cases
- • Aware of cultural conventions (eg. married names)
- • Difficulty with typos, transpositions, nicknames
- • Misses 3% of duplicates
- • Extensive use of fuzzy matching algorithms (e.g. Levenshtein, Metaphone)
- • Explicit management of transitive matches
- • Processing performance becomes a significant issue
- • Misses 2% of duplicates
Information Softworks
- • Embraces inevitable human error (typos, swapped fields, etc.)
- • Able to work around missing data
- • Adaptive and extensible to include any available supplemental disambiguating information (blood type, height, weight, race, ethnicity, etc.)
- • Open source and supported by a suite of analytic tools
- • Unique system automatically identifies and resolves suspicious matches; allows for an aggressive matching threshold without increasing false positive rate
- • Backed by experienced guidance in enterprise deployment
- • Misses less than 1% of duplicates