Errors in diagnosis are relatively common in primary care and many of these errors result in serious harm to patients (Newman-Toker et al, 2013; Auraaen et al, 2018). However, up to 80% of diagnostic errors are considered to be preventable (World Health Organization [WHO], 2016). A mnemonic, SAFER PRACTICES (see Box 1), was created to bring together the most common causes of diagnostic error in primary care as a learning tool, as well as a diagnostic error checklist for use during the consultation (Silverston, 2020a). The emergence of the novel coronavirus COVID-19 has created a new set of problems for clinicians, including an increased risk of diagnostic error. This article describes why the COVID-19 pandemic may result in an increased risk of misdiagnosis in primary care and how SAFER PRACTICES can be used to reduce this risk.
Box 1.SAFER PRACTICES
- S = serious: have I checked whether the patient is seriously ill with COVID-19?
- A = alternative: have I checked for an alternative cause to COVID-19 for this symptom?
- F = fit: have I checked if there are any findings that do not fit with COVID-19?
- E = early: have I checked for an early/atypical presentation of COVID-19?
- R = risks: have I checked for the red flags and risk factors for COVID-19 disease?
- P = problem: was this a problem consultation for me?
- R = re-assessed: is a medical re-assessment of the patient required?
- A = assessment errors: have I assessed the patient objectively?
- C = cognitive errors: have I checked for cognitive errors?
- T = transient reasoning errors: have I checked for transient reasoning errors?
- I = diagnostic isolation: should I check this diagnosis with someone else?
- C = checklist check: have I checked my checklist?
- E = equivocal: is this an equivocal/uncertain diagnosis?
- S = safety netting: have I safety netted the patient?
SAFER PRACTICES is based on a very simple concept which is that diagnosis involves comparing the information gathered during a clinical assessment to the clinician's medical knowledge of symptoms and their causes, through the application of diagnostic reasoning (Silverston, 2020a). This emphasises the fundamental importance of medical knowledge, clinical assessment and diagnostic reasoning to the diagnostic process, and helps to establish why the COVID-19 pandemic increases the risk of misdiagnosis. COVID-19 is a new disease so medical knowledge regarding how the disease presents and progresses, as well as the criteria for diagnosing the disease and determining its severity, is still being formed (NHS, 2020). Without objective criteria to diagnose and determine disease severity, it then becomes difficult to perform a disease-focused clinical assessment and to include the potential diagnosis of COVID-19 within the diagnostic reasoning process. This creates a significant risk of misdiagnosis.
Conventional clinical assessment teaching stresses the importance of gathering information in a comprehensive, holistic and systematic way. This includes gathering information from a global assessment of the patient and their surroundings; detailed history-taking; performing a set of observations; conducting a physical examination; and interpreting the results of tests and investigations. One of the first changes to be implemented during the pandemic was the switch from face-to-face to telephone or video-link consulting. The increased risk of misdiagnosis in telephone consulting can be reduced through training in remote assessment skills, combined with the use of diagnostic algorithms and sound clinical knowledge and experience (Pygall, 2017). However, although it is said that the history forms 90% of the information required for diagnosis, there will inevitably be cases of misdiagnosis due to the inability to gather key diagnostic information from the physical assessment of the patient (Silverston, 2013).
Diagnostic reasoning is a complex process which can be influenced by both internal and external factors, such as changes to working practices, tiredness, stress and mood disorders (Silverston, 2020a). Similarly, the emphasis on confirming or excluding COVID-19 can create abnormal heuristics (mental shortcuts) and cognitive biases within the diagnostic reasoning system, and there is recognition that an increased risk of serious harm to patients exists from not considering other life-threatening conditions during the diagnostic reasoning process (Norman et al, 2017). In response to the increased risk of diagnostic error during the pandemic, SAFER PRACTICES has been modified to help clinicians focus on what they need to know about COVID-19 and how this knowledge can be incorporated into their clinical assessment system and diagnostic reasoning processes.
SAFER
The first part of the mnemonic, SAFER, has two purposes. Firstly, to ensure that the clinician comes to the consultation with the medical knowledge, along with the clinical assessment skills and diagnostic reason processes, required to assess and manage patients with confirmed or suspected COVID-19 disease. Secondly, to act as a diagnostic error checklist during the consultation and also to focus reflection and discussion of cases in which there is diagnostic uncertainty.
In primary care, patients present with a symptom, not a diagnosis, so it is necessary to adopt a symptom-based approach to the consultation. Many of these symptoms are common to both serious and minor illnesses so it is essential to also take a patient-safety focused approach within the consultation (Silverston, 2013). SAFER is specifically designed to bring a symptom-based, patient-safety focused approach to both learning and clinical practice.
S = serious illness
The first priority during the assessment of any patient is to establish whether or not the patient is seriously ill; this is no different in a pandemic. Diagnosing whether or not a patient with COVID-19 is seriously ill or not is complicated by a number of factors. Firstly, COVID-19 is a complex, multi-system disease which can present and progress in different ways (Chen et al, 2020). Diagnostically, it presents a similar challenge to the one posed by invasive meningococcal disease which can present with meningitis, sepsis, mixed disease or atypically. It is also a new disease and much is still being learned about how it presents and progresses so an evidence base for the assessment and management of patients is still lacking and those guidelines that have been produced so far are constantly changing. This makes it difficult to define what constitutes serious illness in COVID-19 disease.
It is important, however, not only to use a trusted guideline to establish the criteria for diagnosing serious disease but also to ensure that the latest guideline is being applied, as these are changing weekly. The National Institute for Health and Care Excellence (2020) has a number of guidelines available, including guidelines on how to identify patients with serious illness in the community (NG163); how to assess and manage patients with specific pre-existing illnesses, such as severe asthma (NG166); and how to assess and manage patients with pneumonia in the community (NG165). There are also a number of clinical assessment and clinical decision-making flow charts available that include information on how to assess disease severity (Barnet Clinical Commissioning Group, 2020; BMJ Publishing Group, 2020). The criteria for diagnosing serious illness should be incorporated into symptom-specific algorithms and computer templates to ensure that a systematic approach is taken to gathering vital diagnostic information during the clinical assessment and that clinical decision-making is based on the best current medical knowledge available.
A = alternative causes
Many of the symptoms associated with COVID-19 are also common to other equally serious medical conditions. It is essential to remember that the serious illnesses that existed in the community prior to the pandemic are still present and should still be considered as equally important potential causes of the patient's symptoms. Creating a list of the alternative diagnoses for each symptom and embedding this information into symptom-specific algorithms and templates can help to reduce the risk that alternative diagnoses will be overlooked. For example, a patient with a headache and fever could have bacterial meningitis, while a patient with a cough and fever might have a community acquired bacterial pneumonia. Bacterial sepsis may present with fever and shortness of breath, and there are many non-infective causes of breathlessness to consider also, such as heart failure or a pulmonary embolus. It should also be noted that patients with COVID-19 are at increased risk of developing a venous thromboembolism and pulmonary embolus (British Thoracic Society, 2020).
F = findings that do and do not fit
One of the most important and useful exercises that can performed is to create a list of the key findings that do and do not fit with COVID-19 disease for each symptom. Diagnostic algorithms use key diagnostic discriminators to distinguish one disease from another. Therefore, knowing which findings do and do not fit with a disease is vital in establishing the correct diagnosis. These key diagnostic discriminators, whether they be questions within the symptom and medical history, or clinical findings gathered from a set of observations or the physical examination, can be integrated into the clinical assessment and diagnostic reasoning algorithms for that symptom.
E = early/atypical presentations
It has been shown that it is often the early and atypical presentations of serious illnesses that are the missed or misdiagnosed (Newman-Toker et al, 2019). Media campaigns have highlighted the presence of a high fever, a persistent, dry cough and shortness of breath as being the typical symptoms of COVID-19 disease. However, these symptoms may only be present later in the disease. Given the importance of identifying patients with early disease for containing the spread of COVID-19 and for providing disease-specific safety-netting advice, it is necessary to know what the early symptoms and signs are. It is also important to be able to identify patients who are presenting atypically with symptoms such as diarrhoea (Han et al, 2020); conjunctivitis (Wu et al, 2020); loss of taste and smell (Yan et al, 2020); and with central nervous system, peripheral nervous system and skeletal muscle symptoms, including symptoms such as dizziness, headache, impaired consciousness, acute cerebrovascular disease, ataxia, seizures, confusion, delirium, vision impairment, neuropathic pain and muscle pain (Mao et al, 2020). There are also reports of COVID-19 being associated with a number of different rashes and viral exanthems (Chia-Jui, 2020; Recalcati, 2020). Patients with COVID-19 may also present with symptoms of a deep vein thrombosis, or a pulmonary embolus (Davoodi, 2020, British Thoracic Society, 2020).
R = red flags and risk factors
Red flag findings are markers of serious illness so it is essential to be aware of the red flags for COVID-19 disease, as well as the risk factors for developing the serious complications of this disease (Public Health England, 2020). This information should then be incorporated into the relevant clinical assessment algorithms to ensure that the red flags and risk factors are proactively checked for during the consultation.
PRACTICES
The second part of the mnemonic, PRACTICES, draws together the most frequently encountered causes of diagnostic error in primary care to increase awareness of these causes and to act as a checklist to prevent and detect these errors. More detailed discussion about the causes of diagnostic error in primary care and the strategies and interventions that can be taken to reduce the risk of diagnostic error is available (Silverston, 2020a).
P = problem consultations
Problem consultations are ones that involve a higher risk of diagnostic error. In a pandemic, there are many factors that will increase the number and variety of problem consultations, including the switch from face-to-face to remote consulting; the shared care of patients between multiple clinicians and multiple healthcare providers; and the need for infection control and containment of the disease. Identifying these risks enables specific strategies and interventions to be put in place to mitigate risk.
R = re-assessment
Illness is a dynamic process and patients with COVID-19 disease have the potential to deteriorate for a number of different reasons. While some patients may be able to monitor their own condition, others will require a medical re-assessment of their condition. It is important that this clinical decision is based upon objective criteria which should include both medical and social factors. Establishing a set of objective criteria to inform this decision can not only help to reduce stress for those involved in making these decisions but it can also improve patient safety. Symptom scoring systems and assessment tools can be used as part of this decision-making process (Greenhalgh et al, 2020).
A = assessment
Remote triage of the patient with suspected or confirmed COVID-19 disease should follow a set of objective criteria to establish whether the patient is seriously ill or not, and whether admission, or a further face-to-face assessment, is required to exclude an alternative cause for the symptom or assess the disease severity. A traffic light system can be developed with a set of criteria to determine which category the patient should be placed in. Patients assessed as being in the red category should be sent to hospital, while patients in the green category can be advised to self-monitor and given safety netting advice. Patients in the amber category will either need to be re-assessed in a face-to-face consultation or receive a remote medical re-assessment, depending on the criteria set for these decisions. There are a number of helpful resources that can be used for this purpose (Primary Care Pathways, 2020).
C = cognitive errors
Diagnostic reasoning involves the use of a complex information-processing system that can is prone to the development of system and human errors (Royce et al, 2019). System errors, such as faulty heuristics and cognitive biases, are more common during a pandemic as reasoning becomes heavily influenced by the overriding presence of the disease. Incorporating a two-stage diagnosis strategy into the consultation can reduce the risk of misdiagnosis due to reasoning system errors. This involves considering the first diagnosis that is made as being the preliminary and not the final diagnosis. SAFER PRACTICES can then be applied to the preliminary diagnosis to check for findings that do or do not fit with this diagnosis, and to check for factors that may have had an adverse impact on the diagnostic reasoning process before the final diagnosis is made.
T = transient reasoning errors
The risk of human errors is also increased during a pandemic, as tiredness, stress, fear, illness and changes in mental health all have an impact on the ability to think and reason. This is another reason to adopt the two-stage diagnosis strategy.
I = diagnostic isolation
Diagnostic isolation is a common problem in primary care, as GPs and practice nurses tend to work independently of one another, affording nurses few opportunities during the working day to discuss uncertain or challenging diagnoses with their GP colleagues. Isolation is increased further in a pandemic by the creation of patient flow systems where access to GPs may be restricted by infection containment measures. Team meetings and informal interactions in the coffee room may be replaced by virtual meetings, restrictions on the number of people gathering and by the requirement for social distancing as part of the infection control measures. GPs may also be working from home, further reducing access to them to discuss diagnostic dilemmas. This can be countered by setting specific times within the working day when uncertain or challenging diagnoses can be discussed with a lead clinician.
C = check the checklist
Experience from the use of checklists as a strategy to reduce the risk of treatment errors shows how it can work best when they are embedded into the treatment plan, rather than allowing the clinician the option of choosing to use a checklist or not. The same rule applies to diagnostic checklists. SAFER PRACTICES works best when it becomes an integral part of the consultation and is used routinely. A laminated copy of the checklist can also be kept in plain sight as a reminder or added to the computer screen as a prompt to remind healthcare workers to check for diagnostic errors.
E = equivocal diagnosis
Diagnostic uncertainty is common in primary care consultations and has to be managed safely if serious harm to patients is to be avoided (Silverston, 2014a). COVID-19 is a new disease, with diagnostic criteria constantly changing as more becomes known about how it presents and progresses. In addition, the restrictions in performing a comprehensive clinical assessment in most patients will increase diagnostic uncertainty still further. SAFER is designed to help clinicians adopt a patient safety focused approach to the problem of diagnostic uncertainty through confirming or excluding the presence of serious illness, alternative diagnoses and findings that do or do not fit with specific diagnoses; to consider the potential for early or atypical presentations of serious illness; and to proactively check for the red flags and risk factors for serious illness and complications.
S = safety netting
Safety netting advice is an essential strategy for reducing the risk of serious harm due to errors in diagnosis, especially when patients present during the early stages of illness (Silverston, 2020b). It is important that an objective assessment of the appropriateness of sending the patient home, rather than arranging a medical re-assessment of their condition, is performed and the ‘5 Cs’ risk assessment tool can be used for this purpose (Silverston, 2014). The next step is to determine the medical content of the safety netting advice that is to be delivered to ensure that the appropriate advice is delivered to the patient for their symptoms and that this advice is presented in a way that is patient-centred (Silverston, 2016). Safety netting advice sheets for patients with suspected or confirmed COVID-19 disease are required which can either be handed to the patient or relative, or sent by email when consulting remotely.
SAFER PRACTICES at work
The first step in implementing SAFER PRACTICES at work is to review the latest information available on COVID-19 with regard to its typical, atypical and early presentations; how to assess the severity of illness in symptomatic patients; the red flags and risk factors for serious illness; and the findings that do and do not fit with the disease. It is important to keep up-to-date with what is known about this new disease. Therefore, a weekly review of the guidelines is recommended at present. The next step is to create a list of the symptoms associated with COVID-19 and under each symptom heading prepare a list of the ‘must-not-miss’ serious causes of that symptom and the ‘must-always-consider’ alternative causes. The final step in this process is to list the findings that do and do not fit with each of these causes so that a diagnostic grid of specific findings to check for can be created and the findings interpreted in the light of each diagnosis. This will help to reduce the risk of diagnostic errors due to medical knowledge deficits and failure in memory recall, along with errors due to not gathering key information during the clinical assessment.
Diagnostic reasoning errors may be due to either reasoning system errors or human errors. Reasoning system errors, such as faulty heuristics and cognitive biases, can be reduced by individuals reflecting on the way that they perform diagnostic reasoning in terms of the way that they balance intuition and analysis within the diagnostic process. The need to consider COVID-19 as a potential diagnosis during this pandemic must be balanced against the need to always consider the possibility of an alternative, non-COVID-19 diagnoses too, if diagnostic errors are to be avoided. The risk of human errors can be reduced through an increase in self-awareness of the causes of human errors in consulting and by developing systems to counter these risks, such as through patient safety focused training programmes and by embedding diagnostic checklists and algorithms into the diagnostic process. Incorporating a two-stage diagnosis strategy into the diagnostic process reduces the risk of both system and human errors.
Discussion
When clinicians are placed under intense pressure, this increases the risk of both diagnostic and treatment errors. One strategy that is routinely employed to reduce the risk of treatment errors is the use of a treatment error checklist embedded into the treatment plan. Similarly, diagnostic error checklists have been used to reduce the risk of diagnostic errors in the high-pressure environment, such as the Emergency Department (Ely et al, 2011; Graber et al, 2014). SAFER PRACTICES provides practice nurses and other clinicians with a system that can be used to reduce the risk of diagnostic errors in primary care during the COVID-19 pandemic.
KEY POINTS
- In primary care, patients present with a symptom not a disease, so it is essential for safe practice that clinicians are aware of the symptoms associated with COVID-19 disease
- Many symptoms of COVID-19 disease are present in other medical conditions, so it is important to consider alternative diagnoses and not just COVID-19 as the cause of the patient's symptoms
- SAFER PRACTICES is designed to help you come to the consultation with the correct medical knowledge, clinical assessment plans and diagnostic reasoning system to avoid misdiagnosis