A friend of mine is the VMO in Emergency Medicine at RNS Hospital. I gave him the PDF of the recent CEO Sessions and the UQ webpage mentioned here in one of the RAP threads. I am certainly not qualified to make assessments or even comment as this is not my sphere at all. But anyway this is the interaction between him and TK:
Have looked at both web sites. I am highly sceptical. I would need independent analysis of the data from both Indonesia ( very small sample size) and WA. Studies can be highly flawed including exclusion criteria and analysis. . I note with concern the potential option of extra clinical data such as fever being added in. Essentially most COPD is due to cigarettes. Respiratory function tests detect COPD well before any ongoing cough. Also diagnosing pneumonia in the third world is all very well but cost of treatment is what is killing children - they can not afford it (I worked in Uganda at a health clinic and saw the reality of economics). I do not accept the capacity to distinguish pneumonia causation. For instance community acquired versus institutional acquired (eg nursing homes) have differing and in common bacterial causation. We know from studies we need broader spectrum antibiotics for institutional causation or people will die - that data is not in an app cough. Also issues with immunocompromised individuals such as alcoholics, cancer treatments and hiv. All these create expanded potential bacterial causes. Essentially if you had someone tell you there was an app that covered enormous potential complexities over diagnosis and treatment you would be highly concerned.
I passed his reservations to TK and he responded below:
I appreciate your friend's concern. That is why our focus for the last 24 months has been gathering the clinical evidence in well-controlled studies to prove the technology. It is also why we are running a large study at three US hospitals. A few points from the comments:
- The inclusion/exclusion criteria from the Indonesian study, and more details on that study, are available in the published literature. In particular this publication:
http://link.springer.com/article/10.1007/s10439-013-0836-0
- The WA studies have not yet been published, however the inclusion/exclusion criteria are close to the US study, details of which can be found at the NIH's clinicaltrials.gov website:
https://clinicaltrials.gov/show/NCT02973282.
- I am not sure why the concern with adding additional clinical data is such a concern? Clinicians use this data to make their diagnosis, and if it is easily accessible then why shouldn't the algorithms also benefit from this? However, that said, we always have published results on cough alone and results with cough and the additional data.
- Yes, lung function tests do test for COPD (and asthma in older children), however these require spirometry and (typically) an experienced operator. They also require the use of a bronchodilator to test if the underlying symptoms respond. For telehealth, there are obvious issues where patients do not have spirometers and should not be administering bronchodilators. I don't believe that many lung function test labs are colocated within an ED, so these tests add a significant delay to treatment (similar, if not longer than getting a CXR).
- I strongly disagree that cost of treatment is the only factor in the developing world (in fact, diagnosis in resource-poor regions was what the Gates foundation initially funded the team at UQ to do). Yes, treatment costs are important (antibiotics, vaccination), but you have to diagnose the patient before antibiotics should be used - and there is often no access to CXR and workers are typically undertrained. Currently WHO guidelines for field diagnosis (based on breathing rate and fever) significantly over diagnose (highly sensitive but not very specific) which causes significant overuse of antibiotics. There is also the fact that many in these regions just don't reach a health worker until the disease has progressed significantly. We are working with our humanitarian partner on solving these two issues using our technology.
- We are only just beginning in diagnosis of causation. Our initial work has showed promise for separation of bacterial, viral and atypical pneumonia. But it is early days.