I’ve been working on a new tool to compare healthcare facilities mapped in OSM with official data from registration/inspectorate bodies. It’s still a work in progress, but you can have a play with what I’ve done so far at https://osm.mathmos.net/healthcare/. It includes my usual mapping progress comparison maps for each postcode area.
Do let me know if you come across any falsely flagged missing or unexpected items due to things that aren’t already documented in the table on the main tool page. Also, can anyone find any data for private healthcare facilities in Scotland, or public hospitals in Northern Ireland?
You can also access the flagged missing/unexpected OSM objects in Survey Me!. The healthcare data is not enabled by default yet, so you’ll need to scroll down and tick the box for it.
Thanks Russ, I’ll have a look. Presumably it doesn’t include private (non-NHS) facilities. It also looks like it’s only got postcodes for locations, which is a shame. (The CQC data includes URPNs for most locations.) On the other hand, it looks like they have data on GP branch surgeries, whereas the CQC data only has one entry for each practice.
Looking a bit closer, it seems that the CSV files don’t have coordinates, and neither do the raw API calls (that it seems will be replacing them soon). But for an individual record, the API does return a UPRN (from which coordinates can be obtained.) I don’t know if there’s a way to get bulk records with the UPRNs included.
Apparently, the CQC profile for some facilities marked on the map have been archived. At first look, this seems to be due to change of operators/service providers. See this example:
Bourne House Nursing Home used to be operated by London Residential Healthcare Limited, but now it is run by Olympus Opco LTD, and has a new profile (Registered on 31 January 2025).
One or two other facilities from the same road happens to have the same situation. I used their new profile IDs for ref:GB:cqc_location while mapping, but I am guessing this will mark them as unexpected when your data is updated.
The “archived” entries in the CQC data should be filtered out by my tool when the data is imported. The example you give says that it was only archived a couple of days ago, so would have been live in the last copy of the CQC data I fetched. If you’re going to add a ref:GB:cqc_location then use the new/current one, as that’s what will work in the long run. The matching should still work based on location alone, as long as the location from CQC is reasonably accurate.
I’ve made some changes to the back-end processing used by the tool and added additional data for “Branch Surgeries” from the NHS. The should mean most satellite GP surgeries that are part of a larger GP Practice will now appear in the data and reduce the number of “false positive” unmatched OSM objects. I’m also working on a way to allow the removal of manually identified spurious/duplicate records from the data.
Two issues with “false positive” unmatched OSM objects that are still to be resolved:
Some people have added individual hospital departments as amenity=clinic or healthcare=clinic. See e.g. https://osm.mathmos.net/healthcare/progress/PE/#16/52.5840/-0.2805. This doesn’t seem inherently wrong to me, but it does create “unexpected OSM object” warnings in my tool because the departments don’t have separate entries in the data. One solution would be to introduce some tagging so my tool can discount these. What would people think to using clinic=hospital_department for them?
There seems to be a quite a few NHS health centre / clinic type facilities that aren’t present in the CQC data. Presumably they host services run by larger organisations that only have one CQC listing to cover multiple sites. At the moment these additional sites can’t be matched and will appear as “unexpected OSM object” warnings. See e.g. https://osm.mathmos.net/healthcare/progress/IP/#17/52.24037/0.70823 I’m not sure what can be done about this. Does anyone have any ideas?
I added ref:GB:cqc_location=1-8379263850 to Way: Richmond Cheltenham (1041757735) about a month ago and it still seems to be showing as a missing care home at the opposite end of the road.
I haven’t yet added the relevant CQC tag to Michael Dental Care (265071915) but it does appear to be the same as 1-204584282 which isn’t too far away, but looks like it’s above the distance threshold.
OSM objects with social_facility=assisted_living aren’t considered by the tool, on the grounds that this sort of light-touch care wouldn’t be registered with CQC. Looking at the description at Richmond Village Cheltenham - Care Quality Commission , I’d say the site should be either social_facility=group_home or social_facility=nursing_home. Either of those would be picked up and matched by the tool. (If people disagree about social_facility=assisted_living then I’m happy to reconsider.)
Currently the maximum distance is infinite if there’s a matching reference number, 500m if there’s a matching postcode, or 150m otherwise. The CQC locations are usually very accurate, though not in this case. Maybe those numbers need increasing a bit - though that runs the risk of having more false matches.