The very interesting study by Kostopoulou and colleagues in the January issue1 highlights the potential value of patients using symptom checkers and handing the results to their doctor at the start of the consultation. This would get over the current technical challenges of a system automatically producing a differential from more complex cases with multiple symptoms.
The study refers to a naturalistic trial of Isabel, a physician-triggered computerised decision support system (CDSS) that showed that junior paediatricians only sought and examined the system’s advice ‘around 2% of the time’. This study was carried out over 10 years ago in NHS hospitals where access to desktop computers was very poor and the use of mobile devices to access the Internet was almost non-existent. This partly explains the low rate, but the more significant issue relates to the standards set by the senior clinicians. In hospitals where Isabel is used and is easily available, and actively encouraged by senior clinicians setting a standard, we have found that it is accessed in about 10% of cases. CDSS can only be a means to help clinicians practise to a certain standard.
The Kostopoulou study is based on three cases all with just one clinical feature that made it technically easier to generate a differential diagnosis. In reality, many cases would have multiple clinical features that would necessitate a CDSS (such as Isabel) that could handle complex free text queries. The study showed significantly less improvement when the CDSS was provided ‘late’. Readers may be interested to know that when we (Isabel) looked at the impact the use of Isabel had on users from their self-reported views, in 17% of cases the user said they changed their working diagnosis after using Isabel.2
The authors state that randomised CDSS study designs are rare. Readers may, therefore, be interested in two recent such studies carried out by two US medical schools.3,4
- © British Journal of General Practice 2015