Who is the best person to diagnose epilepsy? The answer is probably – the best trained and most experienced person, who has the most time to listen and access to a larger team and the resources to perform tests. Often this person has trained as a neurologist. Where neurologists are the scarcest of resources, can newly-qualified doctors be trained to diagnose epilepsy – and how long would this take? Patterson et al (2013) looked at this issue in Nepal and identified the crucial figure, that twenty structured-interviews were needed to push their diagnostic accuracy from 50% to 90%. But doctors often leave their communities so what about nurses and other health workers who tend to more geographically-stable?
Nurse-led telephone clinics feel like a modern-adjunct to an epilepsy service, but in some rural areas this may be the only appropriate way to give advice. Paul et al. (2013) identified that nurse-led services were agreeable to patients and performed very similarly to neurologists in identifying drug-side effects, but less well when deciding to modify an anti-epilepsy drug schedule. How does this approach compare with traditional, face-to-face consultations? Bahrani et al. (2017) randomised 465 patients to either approach. Telephone reviews were less likely to be lost to follow-up, and face-to-face consultations were longer and more expensive.
Moving beyond the humble telephone – Patterson and colleagues (2017) looked at the performance of an epilepsy diagnosis app in Chhattisgarh, India. A benefit of this approach is that the app could be delivered by non-physician health workers. The misdiagnosis rate was similar between this group and local doctors compared to a neurologist – with the doctors displaying a greater degree of certainty over their diagnoses. An app is simply a smart method for delivering a Bayesian tool, where a number of questions can be used to refine a diagnosis (Patterson et al. 2014). This method has been validated and compared with a gold-standard, and could be used in other health-care settings (Patterson et al. 2015).
This type of approach is very attractive when working in health-care settings where ‘standard tests’ such as EEG and MRI are not always available. We lean on these tools also when classifying epilepsy. Kumar and colleagues (2017) looked at using clinical criteria alone to classify 512 cases and found a 97% inter-observer agreement between the assessors. Only 3.2% of cases changed diagnosis with the later availability of MRI and EEG data.
There has never been a more exciting time for UK-based researchers to look at partnerships outside of our shores. The Global Challenges Research Fund should provide the financial muscle for to boost new and existing collaborations. A partnership need not by definition be mutually beneficial – but imbalanced collaborations miss out on fulfilling their potential. We therefore reflect on this Virtual Special Edition of papers that empower local clinicians in resource-poor settings – whilst also thinking, what could we learn from these initiatives in better-funded health-care settings?
