Paper published on decision making process of GPs in diagnosing Lyme disease

This week, a new paper was published following a study to help understand the decision making process of GPs when they are presented with cases of Lyme disease.

Titled ’Understanding the decision-making process of general practitioners (GPs) presented with non-specific conditions’, the study saw eight GPs from the North West of the UK presented with 11 clinical historic cases from patients displaying varying symptoms of Lyme disease.

It made for an interesting read, but raised some questions and concerns for many about the way GPs are approaching and recording primary care diagnoses of Lyme disease.

The lead author of the paper is John Tulloch, who we interviewed in May as part of Lyme Awareness Month.

The research was funded by the NIHR HPRU in Emerging and Zoonotic Infections at University of Liverpool in partnership with Public Health England (PHE), and in collaboration with Liverpool School of Tropical Medicine (LSTM)

The aim of the study

The aim of the study was to understand the decision-making process of GPs in using read codes when it comes to cases that present non-specific symptoms (symptoms that can sometimes be attributed to more than one condition or illness).

Read codes are used and recorded by GPs to code clinical, diagnostic and therapeutic information about a patient when they come to the see the doctor with a complaint. These can refer to specific conditions, or be more general in nature. This information is then stored in the patient’s health record.

Read codes have become key elements of general practice research, but they have their limitations for this use because of lack of quality data, incomplete usage between clinicians and practices, and inconsistent use across the country.

Because read codes are often used in research that hopes to establish the number of cases of disease such as Lyme that are diagnosed in primary care, it’s important to be able to understand how and when these are used when presented with real clinical cases.

How the study worked

The study participants were presented with 11 real clinical cases of Lyme, though they were not give information on the diagnosis, in order to remove any possible bias. Most cases had vague symptomology that could have easily been attributed to a variety of causes other than Lyme disease. The GPs were told that all the patients were otherwise healthy.

A series of questions were asked by the interviewer, including how they would code their patient (using Read codes), and decisions regarding diagnostics, prescriptions, and referrals. The GPs were encouraged to discuss how they reached their decisions. On completion of the clinical cases, the GPs were provided with resources about Lyme disease.

What did the study find?

Out of the 11 cases, five cases contained Lyme disease on their differential diagnosis (list of possible conditions or diseases that could be causing the symptoms), and in only two cases would a GP select a Lyme disease related Read code. In most cases the list of conditions being considered within the differential diagnosis differed from the list of Read codes they eventually selected.

Below are the cases and their presentations:

The paper talks of how the “need to build a strong clinical evidence base was strengthened by multiple GPs enquiring whether the patient in case 1 had a significant travel history, and that some would perform internet searches or ask colleagues about unusual rashes. Similarly, in case 10, all GPs asked for a better description about the firmness and texture of the tick.”

The paper also discusses how GPs from the study seemed to need a certain level of supportive evidence in order to have certainty and confidence in their diagnosis. Apparently, the majority of GPs thought that case 1 with the EM rash was Lyme disease, but subsequently coded the patient with just a general “rash”. This suggests that without diagnostic certainty such as a blood test (which isn’t needed when a rash is present as per the NICE Guideline), a non-specific presenting symptom will be chosen as the Read code rather than one with a definitive diagnosis.

A presentation of the erythema migrans (bull’s eye) rash (this image was not used in the study)

Several of the GPs noted that they had uncertainty around diagnosing Lyme disease, and that without sufficient evidence, would not code it as such.

‘Lyme disease is a possibility here. But I wouldn’t leap to it without a history of a tick bite.’ 

(GP5, case 11)

As we often discuss, this approach can be a problem, as not everyone notices a tick bite – but they may have had exposure to ticks and this is arguably more important.

Some more quotes from the GPs demonstrate the uncertainty and lack of committing to a diagnosis.

‘I think first, “What is the primary complaint?”, and unless there is something definitive, I tend to code with the primary complaint or not code at all. So for this patient, what is the primary complaint? Numbness, dropping things, or is it his blue hands?’ 

(GP5, case 3)

‘If I can’t diagnose, I will pick the main symptom to code. I will always do this unless I’m almost a hundred per cent positive of the diagnosis. Sometimes, i f I’m really not sure, I will write everything in free text and not code anything.’ 

(GP3, case 2)

The coding and diagnostic behaviour also seemed to be influenced by the GPs personal experience of Lyme disease, as demonstrated in these quotes:

‘I would never write Lyme disease on a patient’s record until I had a positive lab diagnosis. I’m wary because of potential litigation, and I don’t want to cause problems for future doctors treating that patient.’ 


As we know, this above attitude can cause a problem because the blood tests carried out in labs aren’t 100% accurate, meaning that this doctor might be reluctant to treat – or at the very least code the conditions as Lyme – rather than make a clinical diagnosis, as recommended in the Guideline.

‘There are so few [Lyme disease] specialists across the country. A friend of mine has Lyme, so I know the difficulties.’ 


The paper’s conclusion

The paper’s authors come to the conclusion that it’s “probable that either the GPs’ knowledge of the various presentations of Lyme disease was poor, or that the GPs would not consider a differential diagnosis of Lyme disease until they had greater clinical evidence, or that Lyme disease would only enter their list of differential diagnoses through a diagnosis of exclusion.”

They summised that the GPs coding behaviour seemed to be influenced by clinical evidence, “professional integrity and defensive practice, diagnostic uncertainty, patient-sourced health information and beliefs, and personal and professional experience£. They explain that this has implications for electronic health records database research based on diseases with non-specific presentations, as there is a strong likelihood that some GPs will be reluctant to code such diseases without prior diagnostic confirmation.

Our thoughts

The findings of this study were, disappointing but sadly not surprising.

The questions it brings about for us are:

  1. What does it say about the confidence and experience of GPs in diagnosing Lyme? And how should this be addressed?
  2. Where and why do these fears questions around professional integrity and litigation appear?
  3. How can they achieve consistency in the coding of Lyme disease (and indeed other conditions) in order to keep more accurate records of conditions as they are diagnosed?
  4. What does this mean for numbers from studies such as those published earlier this year on diagnosed cases of Lyme in primary care that were based on Read codes? It was already acknowledged within the study that numbers were likely to be underestimated because of lack of access to data, but this suggests the number of cases could be much higher than found in this research
  5. Better education and awareness among the medical professional is key, but how can we ensure that this happens? There is already the RCGP’s Lyme Disease Toolkit and e-learning module, but it’s not mandatory for GPs to undertake. And the workshops delivered as part of the Spotlight Project are not continuing.

One observation is that this study took place with a small number of GPs in the North West. It would be interesting to see the study repeated with a larger number, in an area of the UK where incidence of Lyme disease is higher and GPs therefore might be more familiar and confident with presenting symptoms.

Either way, we feel this study gives further weight to claims that cases of Lyme are higher than officially recorded, and that GPs need more support and confidence in learning about Lyme.

*some excerpts from the study are used directly within this piece without citation

Posted in Blog posts.


  1. I was bitten in August 2020 , walking in the woods, at the time I didn’t realise what the bite was- but a week or so later I read an article in a magazine describing a tick bite. I immediately phoned my GP. She asked for a photo of the bite (Covid restrictions) she said it was very unlikely to be a tick bite and I accepted her diagnosis. A couple of weeks later I started to have mild symptoms headache, low mood and mood swings – very unusual for me, then problems with my eyes – I saw another doctor who arranged for blood tests , these came back – Abnormal – my symptoms have become worse, I’m not getting any support , my GP prescribed anti-biotics but as I understand it this is too late to stop the infection. The anxiety and depression is made worse by the lack of support and I feel that my doctor is avoiding the issue.

  2. Hi Jill,

    We are so sorry to hear about your experience. Antibiotics can still be effective even if your diagnosis is delayed, so we would recommend taking what’s prescribed. This should be a three week course of doxycycline at 200mg a day for 3 weeks, and they can prescribe a further three week course of amoxicillin if you still have symptoms after this. This is as per recommendations in the NICE Guideline, which you can read more about here:

Leave a Reply