Calling BiG

Almost 30% of Covid patients in England readmitted to hospital after discharge – study
Readmission rate for Covid patients 3.5 times greater, and death rate seven times higher, than for other hospital patients

Covid affects the old with comorbidities more than others. We’d therefore expect those hospitalised with Covid to require more hospital care more generally than the general population.

So, is all this adjusted for this?

26 thoughts on “Calling BiG”

  1. Bloke in China (Germany province)

    Noted. Some questions for our resident UK GPs, without which it’s difficult to work out what is happening here:

    What is GPES? What data do you feed into it? I presume it captures some data about almost all ambulant primary care patients in England (the preprint claims 56 million individuals), not specifically hospital admissions.

    The control group was taken from people with a “GPES Pandemic Planning and Research” record between 1 Jan 2019 and 30 Sep 2020. From what I can see, the “PPR” part of this is recent. Why would there be any records at all before the start of the pandemic (~ Feb 2020)?

  2. GPES extraction overview
    The GPES is a generic data extraction service operating between NHS Digital and GPSS that allows NHS Digital to query GP systems for data in the form of specific data extractions (an extraction requirement) to meet the needs of a particular data use case. Examples of existing data extractions are those that provide the basis for GP payments or that are used for health screening for example Diabetic Retinopathy.

    The GPES provides standard mechanisms for controlling and scheduling extractions as well as targeting and controlling practice involvement (Participation). This allows control of the population (Cohort) for which data is extracted as well as, where applicable, GP Data Control of whether the extraction is authorised to take place.

    This COVID-19 planning and research extract is a further extract which has been developed by GPSS and is undertaken by NHS Digital to extract the relevant data for central processing.

    The actual subset of available data that is extracted in each GPES extract is defined by a set of business rules. These rules specify features such as the target cohort of patients, the patients qualifying for extraction, the coded record content for extraction and limitations such as time period cut-offs to be applied to the extracted content.

    The following sections provide an overview of the business rules that specify the actual subset of patient data held by GP systems that will be included in the COVID-19 planning and research extract.

    https://digital.nhs.uk/coronavirus/gpes-data-for-pandemic-planning-and-research/guide-for-analysts-and-users-of-the-data

    So basically data extracted from GP records – with attempts at matching based on characteristics.

    https://www.medrxiv.org/content/10.1101/2021.01.15.21249885v1.full.pdf

    To answer TW’s question: yes. (from a cursory read)

  3. Bloke in China (Germany province)

    Was the matched control group actually followed from a discharge from hospital for a non-covid reason? I don’t think so. If this is true then this rather blows the “rehospitalisation” claim. If you want to show prior emergency hospitalisation for X is more likely to result in rehospitalization then you have to compare to people who have actually been hospitalized for emergency (not elective) treatment, for something else, over a comparable period of time.

    You can show anything is associated with poor relative outcomes, including death, if you compare group A “people who bothered a doctor about trivial complaint X” against group B “people who didn’t bother a doctor about trivial complaint X”. The second group invariably contains a lot of people who didn’t bother a doctor about anything, and unsurprisingly bad things are less likely to happen to them. There are a lot of studies like this because they are easy to do, inevitably produce a “positive” result, and the NHS is good at collecting huge databases. It’s an ideal way to get an easy MD, you can do the research in a week or two if you have a decent statistician supporting you.

    I’m not sure they have gone quite this far with this study, but let’s say there are some clarifications that I hope the peer review will add to the paper.

  4. Old/ill as Tim says.And many C19 “infected” are only so because of shite PCR test.

    Lots of conditions–esp serious ones– see people leaving hosp and then going back in. Nature of the condition and/or shite useless NHS didn’t do a proper job the first time.

    More fear porn from the Johnson gang.

  5. Bloke in China (Germany province)

    If the claim is “someone admitted to hospital with covid is more likely to go back into hospital for any reason over a certain period of time than a member of the general public is to end up in hospital for any reason over the same period of time”, then that is totally credible. It is also both obvious and trivial.

    The paper itself illustrates the overcounting you get when looking at people seeking healthcare versus those not. Look at the rate of new onset diabetes in the covid group (1.1%) compared to the matched control group (0.3%). Bear in mind that you already have a vastly higher baseline rate of diabetes in the covid group (24.4%) than the control group (7.0%). Just on the basis that there are fewer patients at risk we would expect to see fewer, not more, new cases of diabetes in the covid group.

    IS there a plausible mechanism by which covid might precipitate diabetes? I don’t know. There is certainly a plausible mechanism by which covid might precipitate CKD, but the difference between groups in new onset CKD (a lot of which is probably also pre-existing disease just being recognised) is much lower (0.6% versus 0.3%).

    By the way, those massive unmatched differences in baseline characteristics (diabetes, CKD, cardiovascular disease, cancer (20.5% vs 9.2%!) dementia (10.5% vs 1.1%) are themselves pretty highly associated with risk of hospitalization. Look at table 1, see how sick the probands are compared to the controls, forget the unknown about whether controls were even hospitalized in the first place, and ask should we be surprised at all to see a high rate of rehospitalization?

  6. Bloke in China (Germany province)

    Ecksy, the PCR test false positive issue creates problems for the simplistic interpretation of the context-free headline numbers that are blasted at us every day. It is totally irrelevant to this study. Patients hospitalized with a primary diagnosis of covid, whether clinical or lab. That decision, that “we gonna stick you in hospital for this” makes the selection, the diagnosis, very (but not 100%) reliable.

  7. Hang on what about hospital acquired covid infections? Does this study take them into account? If patient X is admitted for non-covid health problems, gets covid in the hospital, are they then a ‘Covid patient’ upon discharge? If so, and I believe the NHS has admitted 15% of all covid positive results are hospital acquired, then a good proportion of the ‘covid patient discharges’ have significant other health issues, of a severity that required their hospitalisation in the first place. They must surely be at risk of relapse and requiring more hospitalisation?

  8. How many of those caught Covid in hospital the first time around? Or were they in a care home? Neither places being the natural environment of healthy people.

  9. “I believe the NHS has admitted 15% of all covid positive results are hospital acquired”

    So what is the true proportion? 20%? 25%? 40%?

  10. Bloke in China (Germany province)

    Jim, the group studied was hospitalised with a primary diagnosis of covid, so this is the group of patients that gets seriously ill with it, not hospital-acquired infections. On the face of it, even a hospital-acquired infection that became serious enough in its own right to prolong hospitalisation would not have been captured in this group.

    It is a great case study in how highly selected a cohort can be, even if you use ostensibly simple, transparent, and sensible selection criteria, and how confounded it can be against a control group with extensive matching effort – hence how difficult it is to determine what the outcomes for that group would have been in the absence of the covid infection.

  11. Also, lots of people are currently being sent to hospital when they would usually see their GP. I wonder how this impacts the stats.

  12. Don’t claim to be an expert but wouldn’t a lot of this be explained by sepsis? This is a major feature of covid (2 – 5% of cases according to Global Sepsis alliance)

    “Studies have shown that up to 40% of people who are discharged from hospital after having sepsis will die within two years and at least 60% will be readmitted within one year (Prescott and Angus, 2018; Thompson et al, 2018; Shankar-Hari and Rubenfeld, 2016)” From https://www.nursingtimes.net/roles/hospital-nurses/post-sepsis-syndrome-overview-of-a-relatively-new-diagnosis-08-07-2019/

    I am reasonably sure someone quoted 30% are dead within a year.

    I haven’t checked if the complications are aligned but it seems reasonable to expect a high rate of readmissions and deaths due to sepsis

  13. What they really mean is “almost 30% of very old sick people who’ve been in hospital end up in hospital again within a few months, and sadly some of them die”.

    To which the only rational answer is “Yes, and?”

    But of course there’s been nothing rational about any of this for several months now.

  14. @BiG

    Having perused the paper more carefully, I defer to your judgment on its accuracy, but I think you’ll find that Table 1 is not the control group “Bear in mind that you already have a vastly higher baseline rate of diabetes in the covid group (24.4%) than the control group (7.0%)” instead the paper claims that the control group precisely matches all characteristics in supplementary table 1, which means that the percentage of diabetics in the control group should be 24.4%. They don’t match the sample on the basis of whether they were in hospital in the past year though.

  15. Bloke in China (Germany province)

    Peter, it’d be useful to know if that number is more or less than we would expect. The study seems to be trying to answer that question but I don’t think it does because the control group is not adequately matched on 4 levels: discharge from hospital as baseline (yes for probands unclear and probably not for control), ideally following emergency admission for something comparable, the presence of comorbidities (far higher in proband than control), and comparing 2020 proband data (when people were avoiding hospitals like the plague) to 2019 control data.

    And to belatedly answer Tim’s initial question, the study is well matched (so does not need adjusting) for age, but is not well matched for comorbs (so would need adjusting for, which would need robust previous data, from 2019 or earlier….)

  16. @BiG, but the underlying problem that those who ended up in hospital may have been more ill to begin with is not addressed – co-morbidity factors are being matched, but not whether those who ended up in hospital were more ill to begin with – which was your first point, and a good one. To have confidence in this study, you’d really want to capture all those who were infected including those who did not end up in hospital and look at those with comorbidity factors who did not go into hospital. Of course because the UK (like most places) only tested in hospitals this is not possible.

  17. The actual purpose of the prerevieuw article is to test the waters for the new Gravy Train.

    Betting a beer on Post-CoVid Syndrome(PCS) becoming the new catch-all excuse for the medico’s within a year….

  18. Hmmm… The thing I miss in the article is comparison of rates with hospitalised (past) cases of influenza or pneumonia.

    Both do play merry hell on your body as well when they get that far, and are in many ways similar to serious CoVid cases in side effects.
    Would answer the question of “Is it CoVid?” or “Has your body been given a good Dink, so there’s (inevitable) residual effects?”.

  19. Another excuse to delay lifting lockdowns as the vaccine rollouts reach all of the most vulnerable…aaahhh but there’s this lag of readmission so hospitals still going to be swamped sort of thing

  20. “Betting a beer on Post-CoVid Syndrome(PCS) becoming the new catch-all excuse for the medico’s within a year….”

    The groundwork is already being laid for it. Listening to my local radio station earlier today they was a piece about the local hospital has X patients who had covid and have recovered but still need hospitalisation because of other effects its had on their health.

    Mark my words we are being set up for never ending lockdowns and restrictions regardless of how many people have been vaccinated or not.

  21. On the CoVid front: Sanquin, the dutch blood bank, has released the figures for CoVid antibodies in donors: 13.3% currently, from 10,3% in december. 15% for the two provinces that caught it early and hard.

    https://www.sanquin.nl/binaries/content/gallery/sanquinnl/artikelen/2021/corona_seroprev_sanquin_tm-dec_15jan21.png for purty pictures from the start of the Saga last year without the need for translation.

    The IgM bar shows the % which have had an immune reaction of some kind recently. The CoVid % is specific for the CoVid-SARS-2 antibodies.

    Note that donors here fall in the 11-60 yr old bracket and are entirely volunteers.
    So the vast majority of donors is well outside the Risk category when it comes to CoVid.
    The thing missing, and not traceable because Privacy, is whether or not the positive cases have had symptoms/properly attracted WuFlu. So from this we still can’t judge what fraction of the population that did have it enough to get an immune reaction ran into the virus and didn’t even notice. 🙁

  22. Local health authority announced virtually no influenza community outbreaks this year and concluded that maybe winter measures could be a recurring event

  23. Biggie–as I understand it you go into hosp for something serious and your bogus PCR test says you have covid you go on the list as a covid patient. They may not be treating you for c19 esp if you have no symptoms but that doesn’t mean you are off the C19 list. And if you die for non c19 reasons you will be counted as c19 “victim” even over the protests of yr loved ones.

    And PCR cant tell flu apart. Drs experience maybe able to–but TPTB want flu minimised. Maybe you don’t rock the boat if they TELL you its being classed as Covid. Esp if its an old ill person whose going to die WHATEVER it is. Worth risking yr career over?

Leave a Reply

Your email address will not be published. Required fields are marked *