select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
The NHS has published the latest data on COVID-19 Hospital Activity, including numbers of those in hospital “for” COVID-19 and those in hospital “with” COVID-19 (when COVID-19 wasn’t the primary reason for admission).
Prof Rowland Kao, Chair of Veterinary Epidemiology and Data Science, University of Edinburgh, said:
“While it is important that data be as clear as possible, when looking at the numbers of hospitalisation due to COVID-19, it’s important to remember that the key issue is not the total number in hospital but whether or not COVID-19 was pushing hospitals close to ‘capacity’ i.e. are the numbers sufficiently high that errors become more likely because of staff exhaustion, or that insufficient care is available. These elements of the problem are difficult to quantify, but the front line anecdotal evidence is that, whatever the direct contribution of COVID-19 to persons in hospitals, the NHS services were stretched to their limits over the winter months – and this was under conditions of lockdown where influenza cases were effectively zero – any consideration of the measures in place at the time must therefore reflect both those factors. Adding on to that, that so many hospital patients themselves became infected with COVID highlights the role of in-hospital ‘nosocomial’ infections. Getting infected in hospital is a big problem – and for persons with existing health conditions as would often be the case, covid is much more likely to result in further serious health issues.”
Dr Louise Dyson, Associate Professor in Epidemiology, University of Warwick:
“It has always been very difficult to contextualise hospital admissions numbers because their effect on the NHS depends on so many nuances of the data. For example, 500 admissions in a day means something very different if they are spread around England compared to all being in one hospital. Whether admissions are young people or older people will have an effect on how long they are likely to spend in hospital, and thus on their effect on hospital capacities going forward, and the amount of pressure on the NHS will also depend on the circumstances going into the epidemic wave.
“It is therefore very difficult to place a figure on how many admissions per day is a problem for the NHS. As a result, we tend to contextualise these numbers by looking back at the past – are we going to have “November wave” numbers or “January wave” numbers? Both of these peaks likely had a big effect on the NHS and on the care provided for both COVID and non-COVID patients. It is not really the case that the NHS has a magic number beyond which it is “overwhelmed” and shuts down entirely and under which it is “not overwhelmed” and working perfectly. In the end NHS staff will always do the best they can with what they have, and the consequence of large numbers of admissions is a gradual decline in the level of care that they are able to provide, and the inevitable future effects on non-COVID care as patients have to wait longer for necessary procedures.
“Taking people admitted “with COVID” rather than “due to COVID” out of the admissions figures would then result in lower numbers in the statistics, but it doesn’t make the NHS any less under pressure. The real question is whether the percentage of cases where COVID isn’t relevant has changed over time, as this would make comparisons with previous waves less useful. Of course, the comparison with previous waves is already a somewhat rough and ready estimation of NHS pressures, as the spatial distribution of admissions, circumstances going into the epidemic wave, and the types of cases being admitted will also affect how much pressure hospitals are under.
“In addition, it’s important to note that when patients are primarily admitted for other reasons, but test positive for COVID, this also makes admission and treatment more complicated as they have to be separated from non-COVID patients and treated in a safe manner.”
Prof Karl Friston, Professor of Imaging Neuroscience, UCL, and panellist on the ‘Independent SAGE’ with special responsibility for modelling, said:
“Government advisers and other experts – including ourselves – have been modelling notification rates, certified deaths, deaths with a positive PCR test within 28 days, hospital admission and occupancy data, et cetera on a weekly basis for over a year. The issue of whether people dying from COVID 19, or dying with a positive PCR test – and the issue of whether people are hospitalised due to COVID 19, or hospitalised with a positive PCR test – does not confound estimates of underlying incidence, prevalence and subsequent mortality: one simply includes the reporting criteria in the model (e.g. combining the probability of dying with the probability of testing PCR positive to explain the discrepancy between certified deaths and deaths following a positive PCR test). This has two implications: first, people can probably discount concerns that the government have been using misleading data. Data are data: they can only mislead when improperly modelled or interpreted. And that, to my knowledge, has not happened. Second, it is not necessary to change the reporting criteria, provided one models the way that data are generated appropriately.”
Prof Sir David Spiegelhalter, Chair, Winton Centre for Risk and Evidence Communication, University of Cambridge, said:
“The main role for hospitalisation statistics is to indicate the pressure on the NHS. Patients with Covid have to be treated in a resource-intensive way, whether Covid was the primary reason for their admission or not, and even if they caught it in hospital. Therefore the total number in hospital with Covid seems an appropriate overall summary statistic, although this new breakdown does provide additional information.”
Prof Philip Bath, Stroke Association Professor of Stroke Medicine, University of Nottingham, said:
“When it comes to data on patients hospitalised due to COVID or those admitted with another condition but also happen to have COVID, there is nothing new here! Back in April last year, I was seeing stroke patients who coincidentally had COIVD, and stroke due to COVID. We all knew this back then. There is no deception here – it is simply the way the data are collected. But let’s not forget that from the staff perspective (PPE etc), it does not matter why the patient is there and has COVID – they are still a risk for infection to staff and other patients and extra precautions are still required (e.g. isolation in scarce side rooms) which then slow all hospital processes.”
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“I agree that it would have been interesting and potentially useful to have these figures before now. What I’m much less sure about is how, if at all, it might have changed decisions if we’d had the figures earlier.
“The roadmap test most relevant to this, I’d say, is “Infection rates do not risk a surge in hospitalisations which would put unsustainable pressure on the NHS”. This is broadly about any pressures on the NHS stemming from Covid-19 infections. Someone admitted with a broken leg who turns out, on testing, to be infected will still need to be treated in a way that uses many more hospital resources than a person with a broken leg who is not infected, because of the necessary infection control measures. So infected people who are in hospital mainly for something other than Covid-19 still put extra pressure on the NHS.
“Most of the discussion about hospital admissions and people in hospital with Covid-19 have, rightly, concentrated on trends up and down in the numbers, and on how the pattern of relationship between hospitalisations and new cases has changed over time with the spread of vaccinations. Those trends, measured as rates of increase or decrease, and those patterns, would have been the same if we’d been looking at numbers in hospital primarily to treat Covid-19 instead of numbers who have a Covid-19 infections. That’s because, according to the newly released data, the proportion of all hospitalised people with Covid-19 who are there primarily to treat Covid-19 has changed very little over the timespan of the data. They make up a bit over three-quarters of the total. If the total goes up or down by, say, 10%, then three-quarters of the total will also go up or down by 10% – that’s how these trend calculations work. Those who are saying that using the counts of people being primarily treated for Covid-19 would have changed decisions need to be specific on how the decisions should have changed. If they don’t do that, it won’t be clear how to use the extra data, now that we’ve got them.”
https://www.england.nhs.uk/statistics/statistical-work-areas/covid-19-hospital-activity/
All our previous output on this subject can be seen at this weblink:
www.sciencemediacentre.org/tag/covid-19
Declared interests
Dr Louise Dyson: “Member of SPI-M.”
Prof Sir David Spiegelhalter: “Non-Executive Director of the UK Statistics Authority”.
Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee. I am also a member of the Public Data Advisory Group, which provides expert advice to the Cabinet Office on aspects of public understanding of data during the pandemic. My quote above is in my capacity as an independent professional statistician.”
None others received.