The included data sets consisted of 55 individual interviews and transcripts of 14 focus groups (with 171 participants in total) comprising approximately 1450 pages of text. It was evident that the shift towards remote assessment at the start of the pandemic (driven primarily by infection control considerations) was accompanied by a strong pressure on patients to ‘stay home’ and on both staff and patients to ‘protect the NHS’. Facing a novel and deadly disease, they were encouraged to act in a way that helped reduce pressure on an overloaded system with insufficient staff, tools, processes and systems to provide safe primary care, especially for vulnerable patients. Furthermore, patients and clinicians observed a shift from a more or less holistic and adaptive approach in face-to-face consultations to a more transactional (algorithmic, task-oriented) approach when consulting remotely, raising further safety concerns. We consider these themes in more detail below. Participants in member checking broadly agreed with the findings.

A. One of my main worries about how the pandemic was handled so far is that they’re so scared that people will be overwhelmed, lots of people didn’t seek help at all because they were told not to or didn’t seek help until too late. I think we, we should be erring on the side of assessing people and assessing people face to face to necessarily, you know, if they might need it and I know we’re trying to find a way that, that makes as little waste as possible but we need to be safe and that’s what’s, something that’s been, that’s been missed throughout all of this and I, I just feel that lots of people have died because they didn’t have any help at all. Sorry, I’m quite upset about it actually. (Focus group, GP, data set D, RFG2)

B. We’re dealing with a disease we haven’t seen before (…) It’s all very well having 40 years’ experience of the health service and knowing what a septic patient presents as but this is quite different it can be, it can catch you out, patients can look well and go off very quickly, they can be unwell when they look well. (Focus group, GP, data set D, RFG2 R10)

C. The difficulty is that we have nothing to do to improve things (we are not waiting for antibiotics etc to work), so I am not sure how much people have to deteriorate before they do get admitted. (Delphi, GP, data set C, ID 11567590814)

D. It can be quite hard to differentiate between anxiety and an unwell patient over video especially, and paleness and tachycardia makes me concerned regarding shock. (Delphi, GP, data set C, ID 11576771861)

E. You’re going to have non-COVID diagnoses which have been misattributed to COVID. So, perhaps someone who has heart failure or someone who has had an MI at home and is then short of breath, or you know, asthma or whatever. I mean literally all of the pathology that existed through COVID-19, but which may trigger a clinician to suspect COVID-19 and therefore encourage the patient to stay at home. So, I think that’s one category of error. But the second category is known complications of COVID-19 not being detected, so things like myocarditis, pulmonary embolism, secondary bacterial pneumonia… (Interview, patient/clinician, data set A, KT1)

F. Because of the lack of ability to score her vital signs, I would be concerned I have ‘underscored’ her. (Delphi, GP, data set C, ID 11564537389)

G. In our surgery we have developed a ‘click and collect’ service of a thermometer and a pulse oximeter where patients collect a box with these in from the surgery car park / get dropped of on doorstep in a no contact manner so that we can get them to do the readings at home. (Delphi, GP, data set D, ID 11565076687)

H. I tried to get in touch with my GP and he just said: ‘oh you’ve got your own monitor so you’re fine’. And I suppose there was that niggling thing at the back of my mind ‘well I am sure I’m capable of using it but what happens if I’m not?’ It would have been useful maybe to just have a bit of a conversation around it or, you know, I guess ideally I would have loved for a doctor to have put a finger monitor on my finger and just to give me that reassurance that I was doing it right I guess. (Focus group, patient, data set B, PFG1 R5)

I. The doctor at the COVID hub was brilliant. It was a lady and she was really fantastic and to be honest I think just the reassurance that she gave me was probably all I needed. I wasn’t keen to go in by any means, I did not want to go anywhere near the hospital unless it was absolutely 100 per cent necessary but just the fact that she was talking through the symptoms and, ‘Yes okay that’s fine, just keep an eye on it. If this gets worse or that gets worse please call back,’ so I felt like I had the support there if I needed it and, and that kind of reassured me. (Interview, patient, data set A, IH1)

J. A whole social history is needed – who’s, who’s around, who’s looking after them – because if the people (…) are at risk and there’s nobody there watching over them you have to have a completely different threshold, either having somebody go out and see them face to face or having a paramedic go there for actually getting them into hospital. You’ve gotta do something different (…) depending if they’re on their own or not or, depending who’s with them and whether they’re competent. And that applies to the whole of GPs but I think even more with this because they can deteriorate so quickly. (Focus group, GP, data set D, RFG2 R2)