CQC ratings published in September 2026

This edition covers 589 assessments published in September 2026, relating to 586 locations. It is the first of a monthly series.

Two limits on scope belong up front rather than in a footnote. Collection began on 3 September, and the first run reached back two days to the afternoon of 1 September, so assessments published earlier on 1 September are not all in the dataset. Everything from 2 September onward is covered. So this is not a complete census of the month, and nothing below should be read as one.

The second limit is what the dates mean. These are publication dates, not inspection dates. An assessment appears on the register some weeks after the site visit, so September's publications describe work done across a longer and unevenly distributed period.

There is no comparison in this edition, because there is no earlier month with usable coverage. September is the first month the dataset covers at all, and October will be the first it covers from the first day. Month-on-month figures start with the October edition.

The month in one table

Roughly three in four published assessments were Good. Just under one in five were Requires improvement. Outstanding and Inadequate together account for 44 of the 589, so both ends of the scale are thin, and small movements in either will look dramatic in percentage terms for several months yet.

By sector

Adult social care is 407 of the 589, so it dominates any monthly figure and will keep doing so. Dental practices do not appear at all, and that is correct rather than missing: CQC inspects dental practices but does not award them ratings.

The 48 in the fourth row are services whose CQC service-type strings are not yet covered by the sector grouping used here. That group is known to include independent secondary care, aesthetic and cosmetic services, and termination of pregnancy services. They are shown rather than dropped, because a sector table that quietly excluded 8.1% of the records would read as more precise than it is.

The difference that does not show in the averages

General practice looks like the stronger sector on a first read: 85.7% Good against 70.0% in adult social care. That framing is misleading, because the share rated Inadequate is almost identical. It is 2.7% in adult social care and 2.7% in general practice.

What differs is the spread. Adult social care was rated Outstanding at 5.9%, more than three times general practice at 1.8%, and Requires improvement at 21.4% against 9.8%. Taking both ends together, 8.6% of adult social care assessments landed at one extreme or the other, against 4.5% in general practice.

So these are two different distributions, not one sector performing better than the other. General practice clusters tightly on Good. Adult social care spreads out in both directions, producing more of the best outcomes and more of the middling ones from the same month of publications. That matters for anyone reading a single rating as a proxy for sector quality, and it is the sort of thing a headline average hides.

By region

The gap between the top and bottom of that last column is wide. 29.7% of East Midlands publications were below Good, against 11.2% in London.

That gap should not be read as a statement about care in those regions, for a reason this edition cannot resolve. Sectors are not evenly distributed across England, and the two sectors here have very different shapes, so a region with proportionally more general practice publications in a given month will show fewer below-Good ratings without anything being different about the services in it. One month cannot separate the geography from the sector mix. The October edition can, by holding sector constant within each region.

The Inadequate column deserves the same caution for a simpler reason. There were 16 Inadequate ratings among the 589, 10 of them in the North West and the West Midlands. At that size, three locations moving either way would redraw the table.

The 28 rated Outstanding

Outstanding remains rare, at 4.8% of the assessments collected. 24 of the 28 were adult social care, which is a larger share than adult social care's 69.1% of all publications.

One pattern in that list is worth putting on the record, with a clear caveat about how it was derived. At least nine of the 24 adult social care locations are home care or domiciliary services rather than residential settings, and four national home care networks appear between them: Home Instead three times, and Right At Home, Radfield Home Care and Bluebird Care once each.

The caveat matters. That split is read from location names, because the dataset does not carry a service type finer than the sector grouping. Treat it as an observation to test over several months rather than a measured figure. If it holds, it is interesting: home care is the part of adult social care where the evidence of good outcomes is hardest to assemble, because the care happens in other people's homes with no one else in the room.

What this does not show

• No trend. One month, so nothing here says anything is rising or falling. Comparison starts next month.
• Not a complete month. Collection began on 3 September and reached back to the afternoon of 1 September, so assessments published earlier on 1 September are not all here. The 589 is what was collected, not a census.
• Publication, not inspection. The dates are publication dates. The underlying assessments span a longer and unevenly distributed period.
• Two thin columns. 28 Outstanding and 16 Inadequate are small numbers. Percentage changes in either will overstate what has happened for some months.
• Regional differences are not isolated. Sector mix varies by region and is not controlled for here.
• An incomplete sector grouping. 48 assessments, 8.1% of the records, sit outside it.

Method

Assessments published under the Single Assessment Framework are collected daily from the regulator's own change feed and recorded once per location and publication date, so a location resurfacing in the feed cannot be counted twice.

Collection ran on every day from 3 September to the end of the month, with no missing day. It did not run before 3 September. The feed is queried for locations whose record changed inside a window, at second precision, and the first run's window opened on the afternoon of 1 September. An assessment published earlier that day changed the record before the window opened, so it was only picked up if that location happened to change again later for some other reason. 1 September is therefore partly covered and was never backfilled, and 2 September onward is complete.

Two days from 3 September onward recorded nothing, 13 and 20 September, and both were Sundays. All four Sundays in the month produced either nothing or single figures, against a weekday average of 30 records.

The 589 assessments relate to 586 distinct locations. The difference of three is assessments for a location that already appears, because a location can have more than one assessment published inside a single month. The analysis counts 586 distinct locations, not 586 distinct provider organisations.

A rating and a region are present on every record. 48 records were outside the current sector mapping.

Figures were reconciled before publication: the rating, sector and region breakdowns each total 589, and the 28 Outstanding rows reconcile to every cell in both the sector and the region table.

Last reviewed 2 October 2026