A patient with four records has no allergy history, no visit history and no outstanding balance, because all four exist somewhere else.
Where duplicates come from
| Cause | What happens | What prevents it |
|---|---|---|
| Phone entered differently | Leading zero, country code, spaces | Normalise before matching, not after |
| Name spelled differently | Transliteration varies between desks | Fuzzy match on name plus date of birth |
| No search before create | Desk creates rather than looks | Search-first registration, create as the exception |
| Branch-scoped numbering | Each site issues its own UHID | One identity space across the group |
The clinical cost and the financial cost are the same event
A duplicate record loses the allergy noted at another branch and loses the unpaid balance from a visit last month. Groups usually notice the second first, which is unfortunate, because the first is the one that matters.
Merging is harder than preventing
Every duplicate created has to be merged later, and merging is a clinical act: two histories are being asserted to belong to one person. Prevention at registration costs seconds. Reconciliation afterwards costs judgement, and carries a risk that prevention does not.
Questions we get asked
Why do hospital groups end up with duplicate patient records?
Because registration creates rather than searches, and because the search that is offered fails on the variations that actually occur — a phone number entered with a country code at one branch and without it at another, or a name transliterated two ways. Branch-scoped numbering makes it structural rather than accidental.
Is it better to prevent duplicates or merge them later?
Prevent. A merge asserts that two clinical histories belong to one person, which is a judgement with real risk attached, and it has to be made by someone qualified to make it. Search-first registration with normalised phone matching costs seconds and removes the need for that judgement entirely.
On the figures in this piece. Every rupee amount above is arithmetic on the assumptions stated beside it, not a market statistic. Where you need published Indian claim data with its source named, that sits on claim data. We publish no figure we cannot show the working for.