One of the things that can completely ruin your day when you’re in the business of breaking news is the sheer tyranny of time. While sometimes something that happened a few months ago can still be news – if the events were significant enough and nobody knew about them until you broke the story – too often that lack of immediacy not only lessens the impact, it also renders the event meaningless.
That, regrettably, seems to be the case with Mark Metherell’s piece in the Sydney Morning Herald and The Age this morning headlined “Patient ID glitches dog ehealth system”.
On the face of it, Metherell’s story was pretty troubling: “Glitches in patient identity details for the new e-health system have been found in about one-third of cases nationally, according to a report the Federal Health Department refuses to release.”
It’s the second paragraph where the “news” starts to fall apart: “The secret report shows patient identity information held by state public hospitals frequently fails to match the data that Medicare holds on the same individuals. Differences in the spelling of names or other variations can pose a significant obstacle, as the system requires an exact match before individuals can get e-health access.”
For one thing, since admissions to state public hospitals don’t make up anywhere near one third of national health cases – we’d be a terribly ill population if they did – it cannot be true that mismatches in IHI numbers are running at about 30 per cent nationally, even if mismatches in state hospital records were at that level. And my inquiries suggest that that they are not.
The story appears to have been based on a report which IBM did as part of NEHTA’s DaCS (Data Cleaning Service) Program, which was launched in 2010 to assist state health departments for the Health Identifiers Service and PCEHR.
That program analysed patient data quality, IHI (Individual Health Identifiers) matching and integration with the HI Service, as well as making recommendations on data quality and cleansing initiatives for ehealth adoption. As everybody involved expected, given that public hospitals have not been altogether diligent in relating their patient data to Medicare numbers – it indicated that the hospital data left a good deal to be desired when it came to procuring HI matches.
But the fact is, the report does not show “glitches in patient identity details for the new ehealth system”. What it indicates is data management and data accuracy issues, none of which surprise anyone who has been involved in the area of ehealth for more than five minutes.
And while the report might be “secret”, in fact it’s more than a year old, and the information has already been used to adjust the search strings used in HI matching, so things have progressed significantly.
What the report revealed, essentially, was that batch HI matching using public hospitals’ historical data was not a very good idea. Matching is significantly improved, however, when patient data is uploaded on an individual basis, and particularly if the patient’s Medicare number is verified at the time a shared health summary is uploaded.
I’m told that procedure is producing a matching rate of as high as 99.9 per cent out in the wild.