select search filters
roundups & rapid reactions
before the headlines
Fiona fox's blog

expert reaction to study looking at risk of hospitalisation from the delta variant compared with alpha

A study published in The Lancet Infectious Diseases looks at hospital admission and emergency care attendance risk for SARS-CoV-2 variants of concern, delta (B.1.617.2) and alpha (B.1.1.7).


Dr Julian Tang, Honorary Associate Professor/Clinical Virologist, Respiratory Sciences, University of Leicester, said:

“This study suffers from the same weakness (“Finally, there were no available data on comorbidities, which are known to contribute to hospitalisation risk.24 This study instead indirectly accounted for comorbidities using related covariates, including age, sex, ethnicity, and deprivation.26”) as some of the earlier genomic/modelling-only based studies on clinical severity for the alpha variant:, which did not account for individual comorbidities that can, alone, affect clinical outcomes in the colder months – as well as with COVID-19.

“These later studies on the alpha using individual patient comorbidity patterns actually found no increased clinical severity: and

“Similar individual patient analyses are needed for the delta variant – but because the delta variant emerged during warmer months, the impact of these comorbidities will be less than during the colder months when the alpha variant analyses were performed – so more of the hospitalisations may be due to the delta variant.”


Prof Sheila Bird, Formerly Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:

Points of note are the following: selection of PCR-tests for whole-genome-sequencing is non-random even within the preferential selection of PCR-tests with ct-value of under 30 (a proxy for higher viral load). When double-vaccination fails to prevent infection, it may nonetheless serve to reduce viral load, an attribute that has been claimed in respect of the alpha-variant but not the delta.

“Hence, England’s selection of PCR positives for whole-genome-sequencing is non-neutral by dint of ct-value under 30, even before other confounding factors come into play.

“Surprisingly, therefore, Twohig et al. chose not to replicate Scotland’s analysis, which would have been of great service to science and enabling of meta-analysis.

“A further genomic-selection-bias arises if SARS-CoV-2 infected persons are subsequently hospitalized and undergo a later PCR-test which qualifies for whole-genome-sequencing when the initial PCR-test did not. Rules for admissibility have been devised by the authors which are based in inter-swab-date interval. My concern relates to which PCR-test-date is used at analysis since I have to survive long enough to be hospitalized and, unless hospitalized, I’d be unlikely to have a second proximal PCR-test. Authors’ Table 2 “hospitalization within 14 days of specimen date” lacks clarity when two dates are in contention. Scotland’s approach escapes this and other conundra.

“Linkage is exact, not probabilistic, and relied on NHS number, which was missing for 11% of nearly 50,000 sequence-confirmed cases in England (29 March to 23 May 2021). Even so, linkages were achieved for over 43,000 cases.

“Registration-delays bedevil linkage-analyses. For the most part, Twohig et al. do not analyse deaths (other than as censoring), the registration of which is slower in England, wales and Northern Ireland than in Scotland. However, unlike the Emergency Care Data Set (ECDS), analysts have had to devise definitional work-arounds due to registration delays of up to 8 weeks when linking to the Secondary Uses Service (SUS).

“The authors recognise a host of potential confounders which together give rise to: Age-group (9 categories) * Ethnicity (4) * calendar-week (8) * Lower tier local authority (314) or regions (9) & vaccination-status (2). Indeed, so many cross-classifications [5,184 even with region vs lower tier local authority] that many must be empty. Authors indicate that the number of their strata was fewer than the number of cross-classifications but give scant detail. Wide confidence intervals are the price of over-stratification. In addition to stratification, authors’ regressions fitted linear age, linear calendar week, sex, index of multiple deprivation, and international traveller status. I hope that Schoenfeld residuals were up to the task of assessing goodness of fit (details in online appendix which I have not yet read).

“Adjustment clearly mattered as unadjusted hazard of hospital admission within 14 days of “specimen data” was identical for delta vs alpha but, on adjustment, shifted to be 2.3 greater for delta vs alpha (95% CI: 1.3 to 3.9). Scotland’s adjusted hazard ratio on many fewer cases was 1.9 (95% CI: 1.4 to 2.5). Unlike EAVE II, the English study did not report by vaccine-type, see


Dr Peter English, Retired Consultant in Communicable Disease Control, Former Editor of Vaccines in Practice, Immediate past Chair of the BMA Public Health Medicine Committee, said:

“This paper compares people who were identified, through whole genome sequencing, as having been infected with either the alpha (B.1.1.7) variant (first identified in Kent, and prevalent at the beginning of 2021) or the delta (B.1.617.2) variant first identified in India, which has become prevalent in the UK more recently. The study used various datasets to identify which of these patients had subsequently required emergency care or hospital admission. It found, consistent with previous studies, that the delta variant does seem to be more virulent – more likely to make people ill enough to require such care.

“The paper is from eminent authors from highly respected institutions, and the study appears to have been undertaken very rigorously. The only question I didn’t see an answer to (and I suspect that if I had had more time to study the paper and appendices more carefully I might have found it) was any consideration of possible bias based on whose samples were selected to have whole genome sequencing performed. If some were “symptomatic” patients (and clearly, from the text, some were asymptomatic), then might differences in the symptom patterns and presentation of infections with the different variants have influenced the outcomes?

“The paper confirms the efficacy of vaccination at preventing serious outcomes including those sought in this study – the number of vaccinated patients in either group was small. Having confirmed what we already thought – that the delta variant is more likely to cause serious disease (and given us much more confidence in saying so), what are the implications of this paper?

“We already know that the delta variant is much more infectious or “transmissible” than the alpha variant. I would be very interested in seeing data from studies on transmissions: before delta, we believed that the risks of transmission outdoors were very low. I am not sure that this is still the case.

“The consequence of the delta variant being so much more transmissible is that, if you are not (sufficiently) immune, you are much more likely to catch it. The increased risks of hospitalisation are therefore increased both by its increased transmissibility, and its increased virulence. Of the two, I think that the increased transmissibility is likely to be more important; but of course, the risks add up, and the increased virulence further adds to the importance of avoiding catching the virus if you can.

“Which brings us back to public policies, including the “road map” for returning to normal. I have expressed my concern at the cessation of legally enforceable mitigations, and the effect this has had on behaviours. Many people seem to think the pandemic is over in the UK, whereas the arrival of the delta variant actually means that the pandemic has stepped up a gear. The drop in case numbers after ceasing restrictions took me by surprise. I now think it was probably due to a number of factors including the end of an upward blip in case numbers caused by the Euro football championships (probably mostly due to people gathering in crowded indoor spaces such as bars to watch the match with others); schools breaking up, so that children stopped catching the virus at school and bringing it home to their households; and an artefactual drop caused by there being fewer tests done, because of the holidays. Case numbers have since recovered and are increasing – they have overtaken the numbers at the peak around the time we dropped the restrictions; and I expect this to accelerate as schools return (with a scandalous lack of mitigations) and people start to socialise indoors more as the weather gets colder.

“I remain horrified that people who have no choice but to use public transport or to go shopping can be put at risk by people who refuse to wear masks. Facemasks are a very minor inconvenience, and considerably reduce the risk of somebody who is (often without knowing it) infectious transmitting the disease to others. Wearing a mask is much less effective at protecting the wearer.

“We already knew that the delta variant is more transmissible, and more likely to infect people who have been vaccinated and leave them infectious (while less likely to be seriously ill). We know with confidence, that the variant is more virulent – more likely to cause severe illness requiring hospitalisation or emergency care. Given that it is this delta variant that is now prevalent, we should return to a strictly enforced mask mandate in indoor public spaces such as shops, public transport, and consider whether the variant’s increased transmissibility should mean more “social distancing” and other restrictions, even outdoors. Schools and workplaces also need far greater mitigation (distancing, ventiliation, etc), now we know how much more dangerous this variant is.”


Dr Zania Stamataki, Viral Immunologist, University of Birmingham, said:

“The delta variant was responsible for the uptick in covid cases this summer in the UK, and many of us have heard of even vaccinated people that became infected. This study measures hospitalisations as a surrogate marker of severe disease, and the findings are clear: the delta variant increases hospitalisations compared to the alpha variant previously prevalent in the UK. The majority of cases (74%) were unvaccinated, and 24% were partially vaccinated, with nearly 2% double-vaccinated hospitalised with covid. Taken together with previous studies showing that delta is 50% more infectious than alpha, evidence mounts that we are dealing with a very dangerous variant. Both vaccine doses are needed for maximum protection.”


Dr David Strain, Senior Clinical Lecturer, University of Exeter, said:

“These data confirm what we are seeing in clinical practice, namely that, in addition to the Delta variant being more infectious than the original or the Alpha variants, it is also causing more severe illness, in populations that previously would have had only mild infections. It highlights the need for a comprehensive vaccine program in younger adults and it clearly demonstrates the pre-conception that they do not get severe covid is no longer true.   

“This is not a surprise, as the two things that make the Delta variant more infectious will also have a role in the disease severity. Firstly the Delta variant produces up to a 1000 times more copies when it is replicating. This is not just pertinent when it comes to transmission between individuals, but also when it comes to spreading the virus throughout the body of the person who is infected. Secondly, the modification to the spike protein – the key, as it were, to the cellular lock – makes it easier for the virus to enter the cell thus making the move from viral carrier to infected person much quicker. This combination of more viral copies and better cellular penetration makes it more likely that the cells, tissues and organs will become overwhelmed before the immune system, particularly that of an unvaccinated individual, has had chance to mount a defence.”


Prof Penny Ward, Independent Pharmaceutical Physician, Visiting Professor in Pharmaceutical Medicine at Kings College, London, said:

How does this fit with the existing evidence?

“The information published today is consistent with data previously reported from Scotland which demonstrated that infection with the delta strain (assessed using a less accurate test method than the genome sequencing used in this study) is associated with an increased risk of hospitalisation when compared to infections caused by the alpha variant. It is also consistent with information suggesting higher viral loads in patients infected with the delta variant, higher viral loads being associated with more severe disease. Three quarters of the infections with both variants occurred in unvaccinated patients but the analysis did not include a specific analysis of the outcomes of infection among unvaccinated vs vaccinated persons: indeed at the time of the December peak of the alpha wave in England, the vast majority of the population, including those at higher risk of more severe disease, were unvaccinated. Now, however, 90% or more of the higher risk population have been vaccinated whereas the vast majority infections with the delta variant are still found among unvaccinated persons, who are more likely to be younger age subjects who are less likely to need hospital care following infection than older age groups. These differences account for the jump in relative hazard of hospitalisation following delta variant infection than alpha variant infection following the proportional hazards analyses adjusted for age and other prognostic features for disease severity,

How concerning is this data for the UK/ globally?

“As the UK is rapidly advancing its vaccination program to include the younger age population, and vaccines appear to retain protection against more severe disease (based on the PHE vaccine effectiveness reports published weekly) this data is less concerning for the UK than it may be for countries where vaccine coverage is significantly less, as New Zealand and Australia are now finding as new outbreaks of infection caused by the delta variant take hold in their largely unvaccinated populations. It is comforting that hospitalisation rates and deaths are not rapidly climbing in the UK notwithstanding the apparently lower effectiveness of the vaccines against infection with the delta variant. The information reiterates the importance of individuals accepting vaccination in order to be protected against more severe outcomes, particularly as increased population mixing is likely to increase transmissions as we head into autumn.

What are the strengths and limitations of this study and how do they impact on how the results should be interpreted?

The strength of the study include the use of genomic sequencing to confirm the variant responsible for the infections detected. The study is a database analysis which is not able to accommodate possible differences in management of infections and criteria for hospitalisation between LAHAs (Local Area Health Authorities). Reassuringly, the number of infections found among vaccinated persons was small – reflecting the effectiveness of vaccines – but this meant that differences in severity of illness associated with the different strain types in vaccinees could not be reliably assessed. In addition, insufficient data was available to assess whether hospitalisation rates in the unvaccinated population might have been affected by other co-morbidities. This additional information would be helpful in assessing potential need for additional antiviral treatment in some patients becoming ill despite vaccination to further reduce potential for adverse outcomes of infection in patients at higher risk of hospitalisation/death.

What can we say about this effect in fully vaccinated individuals?

“As noted earlier the majority of infections were in unvaccinated persons. The number of breakthrough infections found in vaccinated individuals was too low to compare with confidence, but the observed reduction in hospitalisations and deaths observed during the current wave suggests that vaccination while possibly being less protective against infection nonetheless continues to protect infected patients from more severe outcomes. This could perhaps be addressed by careful prospective analysis combining data across the UK as a whole and focusing specifically on the pattern of disease following break through infections found in vaccinees.

Any other comments?

“As noted above additional work is needed to clarify the impact of vaccination on the patterns of illness found in subjects with breakthrough infection as this might inform efficient use of antiviral medications which are just starting to become available.”



‘Hospital admission and emergency care attendance risk for SARS-CoV-2 delta (B.1.617.2) compared with alpha (B.1.1.7) variants of concern: a cohort study’ by Katherine Twohig et al was published The Lancet Infectious Diseases at 23:30 UK time Friday 27th August.



All our previous output on this subject can be seen at this weblink:



Declared interests

Prof Sheila Bird: “SMB is a former programme Leaders at MRC Biostatistics Unit, Cambridge where several co-authors are based but had no input to the study today be Twohig et al. SMb holds Honorary Professorship at Edinburgh University’s College of Medicine and Veterinary Medicine where EAVE II’s first author is. However, SMB also had no input to EAVE II by Sheikh et al.

SMB is also a member of the Royal Statistical Society’s COVID-19 Taskforce which has called for greater transparency in reporting vaccine-status by vaccine-type. SMB lead for the Royal Statistical Society on the need to end the late registration of deaths in England, Wales and Northern Ireland.”

Dr Peter English: “Dr English is on the editorial board of Vaccines Today: an unpaid, voluntary, position. While he is also a member of the BMA’s Public Health Medicine Committee, this comment is made in a personal capacity. Dr English sometimes receives honoraria for acting as a consultant to various vaccine manufacturers, most recently to Seqirus.”

Dr Penny Ward: “No COIs. I am semi-retired, but I am owner/Director of PWG Consulting (Biopharma) Ltd a consulting firm advising companies on drug and device development. Between December 2016 and July 2019 I served as Chief Medical Officer of Virion Biotherapeutics Ltd, a company developing antiviral treatments for respiratory viral diseases. Previous employee of Roche, makers of tocilizumab (anti IL6 antibody) and CMO of Novimmune, makers of empalumab (anti IFN gamma antibody). These are my personal views and do not reflect those of either institution.”

None others received.



in this section

filter RoundUps by year

search by tag