The difficulty of brief medical records and AI

I have previously written about cases where the accuracy of medical records has been brought into question.  In the case of Shaheen v Dr Daish [2025] EWHC3056 we have a case where the court had to determine the likely content of a GP consultation in February 2019 where the note was brief and lacking in information.

Briefly, the case related to allegations of a delay in diagnosis of a lung tumour arising from the failure to carry out a chest x-ray.  Whilst the parties agreed that had an x-ray been done it would have revealed an abnormality that would have led to a CT scan being recommended, it remained in dispute as to whether subsequent treatment would have altered the outcome.

However, the trial was concerned only with facts of events at the GP consultation.

The process for requesting a chest x-ray was that:

• The doctor had to request it on the system.
• The patient had to be informed of the request and be advised to attend a walk-in radiology centre for the x-ray.
• Unless the patient attended, nothing else would happen, as happened in this case.

It was therefore essential that a patient be informed of the request and the process.  This arrangement was not dissimilar to the process for undergoing other investigations that the deceased had had in the past, for example, blood tests.

The limited factual issues to be determined at Trial were:

1. Did the doctor fail to inform the deceased that a chest x-ray was required and that he needed to attend the local walk-in radiology department;
2. If it was the case that the GP had failed to inform him but instead had had done so, would the deceased, as a matter of fact, have attended for his x-ray?

There was a third point relating to contributory negligence, but this article is concerned only with the factual issues relating to the consultation.

The question of the events of the consultation was interesting.  The GP’s note made no reference to the plan to perform a chest x-ray nor did it state anywhere in the records that the deceased had been advised of the plan.

As is often the case, many months or years following a consultation, the parties’ recollections of the events were hazy.

The deceased made a witness statement prior to his death in which he had stated that he was not informed of any need for a chest x-ray at the consultation.

Perhaps unsurprisingly, the GP’s statement and evidence at trial was that she had no specific recollection of the consultation and that her evidence was derived from her review of the medical records and her usual practice.

Her position was that it was a matter of routine to explain the process to the patient and therefore she would have done so.  In her evidence she also said that when informing the patient of the need for the chest x-ray, she would have told him that the purpose was to rule out more serious causes for his symptoms, specifically, cancer.

Interestingly, when the deceased had seen the GP’s colleagues at other consultations, those colleagues had made specific entries in the medical records describing the information that they had provided to the deceased as to the procedure for undergoing various investigations.

At the conclusion of the consultation, the GP had prescribed an inhaler for his increasing shortness of breath.  The plan was for a review in two weeks’ time, which she stated was to provide time for not only the medication to work but also for a chest x-ray to be done and reported.

However, after the deceased left the consultation, the GP stated that she decided also to add in a steroid.  The evidence showed that the surgery had sent a text message to the deceased after he left the surgery advising of this additional step (the prescription of prednisolone) but that there was no such text message in relation to the decision to request a chest x-ray.  This was significant, because it was apparent that the request for the chest x-ray was entered onto the system after the deceased had left the consultation and at broadly the same time as the decision to add in the steroid.

By contrast, the claimant submitted a number of witness statements from various witnesses, none of whom were present at the consultation, but each of whom described the general nature of the deceased and his concern for his health.  Evidence was heard that he was a meticulous businessman who never missed an appointment.  However, this was in contrast to a number of missed medical appointments in his medical record following the Februar 2019 appointment.

It did however become apparent that at the time in question, he was exhibiting some considerable anxiety regarding his health.

The judge considered that, given the evidence surrounding the deceased’s anxiety for his health and that had he been told about the need for an x-ray, he would also have been told that the reason for it was to rule out cancer, the judge felt it highly unlikely that he would have failed to attend for the x-ray or failed to attend for the subsequent follow-up appointment two weeks’ later.  The judge therefore concluded that the most likely account of the consultation was that the deceased was not informed about the need for the chest x-ray, and thus the GP was in breach of her duty of care to him.

It followed, given the nature of the judge’s findings, that had the patient been advised of the need for a chest x-ray and the reasons for that x-ray, he would have attended for the x-ray and attended for follow-up.

The claimant therefore succeeded on those issues of breach of duty of care and initial factual causation.

Comment:

The problem of brief medical records arises frequently in medical negligence cases, particularly where there are short consultations often undertaken under a degree of time pressure.  And establishing the facts is a necessary step to enable the medical experts to offer a view on the standard of care provided.

The Court need to consider all of the evidence and in particular that of the patient and doctor and their responses to questions put under cross-examination to determine what is likely to have happened.

However, with greater uptake of AI tools in medical practice, it may be that trials of this nature will diminish and disappear.

Many are familiar with the extensive deployment of AI tools diagnostic imaging, but what is less widely appreciated is additional areas where AI can impact on medical practice.

“Ambient Scribe Software” describes a set of applications that can listen to a consultation and generate a transcript, provide a summary note, or a list of action points etc.

It seems that uptake is currently slow but progressing.  However, with wider application I suspect that fact finding Trials of this nature may well fall away.  We should be able see more frequently complete contemporaneous accounts of consultations, which remove some of the inevitable guesswork of recollection many years following an appointment.

It is in cases arising from allegations of inadequate consent where there might be greatest impact.