Clearly the down tick at the end is the result of PHE struggling to adjust for the incomplete data from a week with a bank holiday followed by a week with two bank holidays. Paperwork, such as registering deaths, tends not to get done at weekends or on bank holidays.
So, theoretically, we could have had any of the red dashed line corrections:
EDIT We now have the 2021 week 2 report with more delayed registrations included. Looks like all cause mortality is now on a par with 2017/18. Still watching...
What happened when the majority (most, though not all, comes through within a week) of data were added in? On the 7th of January an update was posted so here is the original graph directly alongside the new graph. Again, ignore the downstroke:
If anyone is struggling to see the difference I've circled the important areas in red:
To put the current situation in to perspective I've pulled the curve from the last years during which we had a significant winter flu epidemic, 2017/18:
and just to make it even clearer, here's a line to show peak all cause mortality from 2017/18 vs end of Dec 2020:
It's also worth noting, again on a terribly parochial UK basis, that the winter peak in all cause mortality is almost always at the end of the first week of January. As in now, though the numbers will take a week or two to fully come in. This year may be typical, it may not. So far it's typical.
I have absolutely no doubt whatsoever that work in the ITUs (which are busy but not full, on a national basis) is hell at the moment. In hot-spots, worse. Given the level of barrier nursing needed, staff shortages because people are not allowed back to work until they eventually become PCR-ve (which can take months), social distancing within wards etc life must be awful and exhausting.
I absolutely stand ready to accept that there is a catastrophe ongoing at the moment. It is quite possible that this winter's peak deaths will exceed that of 2017/18.
So far it hasn't.
I would suggest that the Guardian and the BBC may not be the most reliable sources of information about the current situation in the UK.
Another (as so often!) addendum:
Here are the ITU admissions for COVID-19 from ICNARC (which I consider to be unbiased reporting) up until just before Christmas:
which looks pretty grim, despite UK ITUs being at under 100% occupancy. But these are just COVID-19 patients. I don't have a similar curve for non-COVID-19 patients, but I do have this graph which includes both groups (and the rest of England), should they go on to die:
The last column is too low in both death categories due to delayed reporting, as always.
It's quite clear that for every death from COVID-19 there is nearly one less death from anything else. It could be that the virus is so lethal it is impossible to die quickly enough from anything else before it gets you.
Or it could be that managing to get a SARS-CoV-2 PCR-ve death certificate is becoming harder and harder to achieve in a hospital setting when PCRs are still being run to 45 cycles during a winter of endemic COVID-19.