Previous opinion bias describes how the diagnostic process may be influenced by an investigator who is aware of previous test, results and diagnoses in a patient. Having previous opinions has a knock-on effect on the prior probability of a symptom or sign being recorded as present or of a diagnosis being included or excluded.
A doctor is asked to examine an unwell patient by his colleague. How the first doctor makes this request is vital if bias is not to be introduced. If the doctor asks, ‘Could you examine this patients heart, I think he has a murmur’ this introduces a previous opinion bias, as it is difficult to consider that the patient has anything else than a heart murmur. Previous opinion bias occurs if a colleague passes on the diagnosis: ‘could you examine this patient, I think they have a DVT,’ or if an opinion about treatment is passed on: ‘I don’t think this patient will benefit from this treatment, do you?’
The impact of previous opinion bias is hard to determine. It will impact on the decision to undertake subsequent testing, the accuracy of diagnosis, or the outcome or treatment decisions made. In areas such as emergency departments, where there is a need to work at speed and with numerous colleagues previous opinion bias may introduce cognitive errors. Previous opinion bias has many similarities with confirmation bias, where initial or preconceived ideas about something lead to the collection of information that confirms a given view.
To reduce uncertainty, health care professionals often consult colleagues for a second opinion. Minimise previous opinion bias can be reduced by asking open-ended questions: ‘Could you examine this patient – I need a second opinion,’ and by not asking leading questions: ‘Could you confirm that….’.
To reduce biases such as previous opinion bias, every researcher and healthcare practitioner must strive to observe and use the best available information, in the best possible way, being aware that one’s preconceptions can be misleading.