When he originally saw the doctor in Outpatients, she took a full history and then referred him for diagnostic tests. He left thinking she was a good doctor. It wasn’t long before he was back in Outpatients to see the doctor again. ‘The test results confirm that your problem is what I thought last time,’ she said. ‘But before we go any further, please will you tell me again why you originally came to see me.’ It was then that he realised she was a very good doctor.
Correctly diagnosing a disease is a complex task. The same disease can present differently in different patients. And different diseases can have similar presentations. Diagnostic test results can yield false positives and false negatives. Most misguided care results from a cascade of cognitive errors; so, communication is key to making sure that the doctor has not overlooked some key information or considered it to be unimportant1.
There is a strong parallel between the human body and the world of modern digital healthcare, which requires a level of analytics as detailed as that used by the very good doctor2. Just as the doctor takes care to minimise the risk of misdiagnosis, the very good digital healthcare worker has the humility to realise that they too might be wrong. As Adam Grant writes3: A sign of wisdom is thinking in probabilities rather than certainties. Naïve people are arrogant. They claim to know what will definitely happen, and they believe they are always right. Wise people are humble. They explore what might happen and admit when they are often wrong.
Here are three examples of issues faced by different NHS digital healthcare workers and possible approaches to ensure the correct ‘diagnosis’ is made, and the correct ‘treatment’ is delivered.
- Implementing a new computer system
A computer system that works well in one NHS organisation may not easily work well in another. The history of the NHS (part local history and initiatives, part national funding and initiatives4) means that the diabetic service has been run differently in different parts of the country. In one trust the diabetic service has for at least 40 years been part of the hospital service. Thus, the patient’s contacts are managed via the hospital-based Patient Administration System (PAS) and the funding for the service is derived from PAS data (used for contracting and Payment by Results5). Aspects of the patient’s diabetes care are visible to other hospital specialties when the patient is being treated for other conditions – a requirement for safe care. But in the second trust the diabetic service has always sat within the community service, and the patient’s contacts are managed within a ‘community’ computer system with better functionality than PAS for the community clinicians.
Implementation of the ‘community’ system within the hospital service cannot be just a case of ‘plug and play’. It can be done, but not without acknowledging and accounting for the differences between the two trusts. A key question to ask of any system supplier is ‘If we implement your system then how will we need to change the way we work?’ But the full answer to that question – the one needed to ensure a successful and safe implementation – can only really be obtained by the trust’s own implementers repeatedly questioning all the people involved in the delivery of the service now. This includes the appointments officers, all the health professionals who see the patients, the data experts, those who report the data, the finance staff who use the data to fund the service and the directors responsible for patient safety between specialties – and that list is not exhaustive. To try to impose a system currently used in one NHS setting onto a second NHS setting would be to ignore the ‘work-as-imagined’/ ‘work-as-done’ dichotomy6. It is hard work to find out how existing customers really use a system and source the information they record, but it is essential work.
2. Using ICD 10 data to answer a ‘simple’ question.
Hospital doctor to data analyst: ‘How many patients with pressure sores did we have last year?’.
Analyst to coder: ‘What’s the ICD107 code for pressure sores, please?’
Coder to analyst: ‘L89’.
Analyst runs a search for instances of code L89 last year.
Analyst to doctor: ‘There were 129 cases.’
But while L89 is the correct code, if the coder had been asked a different question, or the analyst had remembered that there are coding rules8,9 that analysts should at least understand the existence of, then the coder would have added that L89 followed Y95 (‘nosocomial condition’) means that the sore originated during the stay in hospital; and that L89 not followed by Y95 means that the sore was present when the patient was admitted. Which patient group is the doctor really interested in? Does she understand that the coding rules allow for the two groups to be recorded, searched for and reported separately?
It is natural to think that if we had access to primary care SNOMED CT10 data then this could replace the ICD10 codes which are not recorded until after the patient has been discharged. But these ICD10 and SNOMED CT codes are not interchangeable. Here is the answer to the question ‘Does SNOMED CT remove the need for ICD 11?’11
No. SNOMED CT is a clinical terminology, designed to provide a machine-readable version of the information recorded by clinicians at the point of care in an Electronic Patient Record (EPR). It is a highly expressive terminology that supports a broad range of use cases, including interoperability, decision support and reporting, but it is not designed to be a classification. ICD is the global standard for the classification of diseases and related health conditions and ICD-11 has been purpose built to provide stable, comparable and statistically valid data. By combining SNOMED CT and ICD-11, users can benefit from both rich, machine-processible point of care data and robust statistical data.
3. Communication about patients between different healthcare organizations
For patients waiting for surgery, the anaesthetist and surgeon need to know about any relevant medical conditions (‘comorbidities’) which may affect a patient’s recovery from the anaesthetic and the success of the surgery itself. The hospital pre-operative assessment nurse checks the patient’s GP record for details of any comorbidities. She then also checks the hospital’s own records and finds a recent letter from a doctor to the GP advising that the patient has a possible diagnosis of obstructive sleep apnoea. But this possible comorbidity is not in the GP record (see previous section). Perhaps the practice has been short-staffed and there is a backlog of hospital correspondence to be ‘coded’ onto the GP system, or the practice has a policy of not recording diagnoses until they are confirmed as definite12. For the surgeon and anaesthetist, the ‘possible’ diagnosis may still be relevant. Systems or analyses that look to support integration of care between organisations need to recognise the above reality. Primary care SNOMED CT data lacks an advantage enjoyed by the hospital-based ICD10 data. Implemented in the NHS in 1995 (and preceded by earlier revisions), ICD 10 has been critical to the allocation of £ billions since 2005 (when the NHS implemented Payment by Results5). There are therefore nationally applied recording rules and incentives to get it right; primary care does not have anything equivalent governing the recording of SNOMED CT data.
As a digital healthcare worker, you need to continually deepen your understanding of the healthcare, NHS and clinical processes by working closely with receptionists, care coordinators (e.g. pre-op assessment nurses), clinicians, system administrators, operational managers, clinical coders, etc. to understand what you don’t know, and that there will always be relevant things that you don’t know. And even when you have got the answer to a question you still need to behave as if there may be important facts you still don’t know: does the process just described by the frontline worker or manager apply in 100% of situations, or just most of the time? Why might it not apply and what is the workaround in those situations? Being curious and exercising humility means that your digital healthcare task has a greater probability of success, just like the Outpatient doctor’s task of correctly diagnosing and treating the patient.
Note: The views and opinions expressed in this article are not necessarily those of Northumbria Healthcare NHSFT.
References
- Groopman J, How Doctors Think, p260-261 (Most misguided care results from a cascade of cognitive errors. Different doctors have different styles of practice, different approaches to problems. But all of us are susceptible to the same mistakes in thinking. How to make the correct diagnosis? There is no single script that every doctor or patient should follow. But there are a series of touchstones that help correct errors in thinking. Doctor and patient will start again searching for clues to solve the problem. The first detour away from a correct diagnosis is often caused by miscommunication. So a thinking doctor returns to language. “Tell me the story again as if I’d never heard it – what you felt, how it happened, when it happened.”…Telling the story afresh can help you recall a vital bit of information that you forgot. Telling the story again may help the physician register some clue that was, in fact, said the first time but was overlooked or thought unimportant.)
- David Newey (@NHSNewey). X (formerly Twitter) 15/2/2023
- Adam Grant (@AdamMGrant). X (formerly Twitter) 18/11/2023.
- Beattie A. What if M&S was run like the NHS? Health Service Journal 2018. https://www.hsj.co.uk/service-design/what-if-mands-was-run-like-the-nhs/7023058.article
- https://www.gov.uk/government/publications/simple-guide-to-payment-by-results
- Braithwaite J. Changing how we think about healthcare improvement. BMJ. 2018; 361:k2014. https://doi.org/10.1136/bmj.k2014 (Quoted in Alastair Beattie and Birju Bartoli. Organisational intelligence and successful change in NHS organisations. British Journal of Healthcare Management. 2020;26(3). Available free from AB to all working within the NHS.)
- https://en.wikipedia.org/wiki/ICD-10
- https://classbrowser.nhs.uk/ref_books/ICD-10_2022_5th_Ed_NCCS.pdf
- https://alastairbeattie.org/2023/06/25/the-midwife-icd-10-and-clinical-coding-in-the-nhs/
- SNOMED CT – NHS Digital
- ICD-11 FAQs – Delen: Home – NHS England (kahootz.com)
12. IT failures causing patients deaths, say NHS safety body. https://www.bbc.co.uk/news/health-67503126
The NHS: observations of an Information Specialist 