I thought it would be worthwhile starting a new thread hopefully clarifying some basics for people with little medical or research background and explaining what the ResApp technology potentially can and can't do. I've been quietly reading some of the posts recently and there's a lot of confusion and anger out there and some non-sense written. There are no conspiracies, No-one is out to trick you. It's your money and your investment in new technology. Everyone is entitled to an opinion, but some are worth listening to more than others. I am a medical doctor - for what that is worth - I have been involved in treating patients and medical research for more than 20 years - but I'm no expert in respiratory medicine. I own the stock - I like the long term story. Here is my assessment.
Some facts to digest when looking at results and assessing this new tech and putting it into perspective. I'm focusing on pneumonia here as that was the poorest recent US result. The Aus results will take it to CE mark within the next 3 months.
1- Definitions: Positive percent agreement (PPA = Sensitivity x 100) = The New test is positive and the disease and reference test is positive =True Positive. The New Test is positive, but no disease = False positive. Negative percent agreement (NPA= Specificity x100) = the new test is Negative and no disease = True Negative. The new test is negative, but disease present = False negative. Thereis no perfect diagnostic test – some tests are highly specific – ( high Neg Percent Agreement - NPA)but are only present in a small proportion of cases ( low Pos Percent Agreemnt - PPA ) and viceversa.
2- Mostdiseases are not diagnosed based on one single test – each test has a certainsensitivity (PPA) and specificity (NPA) and often you need a combination of tests to makethe diagnosis. See article below by Dabbagh and the relevant tables I’vecopied.
S. A. Al-Dabbagh and S.N. Al-Zubaidi. The validity of clinical criteria in predicting pneumonia amongchildren under five years of age.
J.Family.Community Med. 11 (1):11-16,2004. ( This gives real insight into current diagnostic methods and showsno one test in isolation gives the answer – often a combination is needed toyield high accuracy – in this study to make the diagnosis of pneumonia youneeded signs on X-ray) – but X-ray itself is only 70% accurate - and open to interpretation)
You will note here the most diagnostic clinical signs whenlistening to the chest are the lower 2 = Wheezes and Crackles – Wheezes – PPA =61% , NPA = 33% and Crackles – PPA = 94%, but lousy NPA = 27% . these arethe relevant numbers to compare to ResApp info. The only 2 signs thatbetter ResApp are Temperature and Grunting.
In telehealth when the doctor can't listen to your chest - the ResApp diagnostic is the only alternative. I think if you asked thequestion {“ Is the temperature greater than 38 degrees?” ( or maybe newercriteria suggest 38.5 degrees ) on the smart phone app as part of the 3 or 4preliminary questions – it shouldn’t be hard to get an even higher accuracy. ( A wheeze and no temp = asthma, high temp and wheeze or crackles = likelylower resp infection / pneumonia. )
3- Regardingpneumonia – it is largely a 3rd world disease with prevalence in the western world of only between 1 and 5 in 1000 cases. the X-ray is only one of a series of tests together with clinical signs thatneed to be positive in order to make the diagnosis and there is stillcontroversy over the role of X-rays. If consolidation is present on X-rayit confirms the diagnosis = high specificity, however if not seen it does notexclude the diagnosis. The WHO definition is copied below: note in 3rd world countries X-ray often not available.
WHO classification ofCommunity Acquired Pneumonia in children – recommendations 2014 – ( note nomention of X-ray diagnosis)
Recommendation 1
Children withfast breathing pneumonia with no chest indrawing or general danger sign shouldbe treated with oral amoxicillin: at least 40mg/kg/dose twice daily (80mg/kg/day) for five days. In areaswith low HIV prevalence, give amoxicillin for three days.
Children withfast-breathing pneumonia who fail on first-line treatment with amoxicillinshould have the option of referral to a facility where there is appropriatesecond-line treatment.
Recommendation 2
Children age2–59 months with chest indrawing pneumonia should be treated with oral amoxicillin:at least 40mg/kg/dose twice daily for five days.
Recommendation 3
Children aged2–59 months with severe pneumonia should be treated with parenteral ampicillin(or penicillin) and gentamicin as a first-line treatment.
— Ampicillin:50 mg/kg, or benzyl penicillin: 50 000 units per kg IM/IV every 6 hours for atleast five days
— Gentamicin: 7.5 mg/kg IM/IV once a day for at least five days
and this Regarding ChestRadiography
From Clinical Practice Guidelines by thePediatric Infectious Diseases Society and the Infectious Diseases Society ofAmerica (2011)
https://academic.oup.com/cid/article/53/7/e25/424286
Initial Chest Radiographs: Outpatient
· 31. Routine chest radiographs are not necessaryfor the confirmation of suspected CAP in patients well enough to be treated inthe outpatient setting (after evaluation in the office, clinic, or emergencydepartment setting).
(strong recommendation; high-quality evidence)
· 32. Chest radiographs, posteroanterior andlateral, should be obtained in patients with suspected or documented hypoxemiaor significant respiratory distress (Table 3) and in those with failed initialantibiotic therapy to verify the presence or absence of complications ofpneumonia, including parapneumonic effusions, necrotizing pneumonia, andpneumothorax.
(strong recommendation; moderate-quality evidence)
4- X-raythemselves are not that reliable from an inter-observer point of view. I couldn’t find any studies on reliability of X-raydiagnosis in children – but this interesting one in adults:
G. W. Waterer. TheDiagnosis of Community-acquired Pneumonia. Do We Need to Take a Big StepBackward?
Am.J.Respir.Crit Care Med. 192 (8):912-913, 2015.
“In this issue of the
Journal, Claessens andcolleagues (pp.
974–982) presenttheir study of 319 patients with a clinical diagnosis of Community Acquired Pneumonia (CAP) who had both achest X-ray and a thoracic computed tomography (CT) scan at the time ofadmission (
2).Disturbingly, 30% of patients who were felt to have CAP based on thepresentation and chest X-ray had no evidence of pneumonia on CT scan.Furthermore, one third of patients who had no change on chest X-ray had CTchanges consistent with pneumonia. Overall, the CT scan results showed that thecombination of clinical features and chest X-ray lead to a misdiagnosis of theabsence or presence of CAP in nearly one-third of all patients studied, andclinicians adjusted their perception of the likelihood of CAP being thediagnosis in more than 50%. If confirmed by further studies, this shifts theassessment of the diagnosis of CAP from “we might occasionally get it wrong” to“Houston, we have a problem.” The implications for everything from empiric therapyto reimbursement and quality of care measures are enormous.”
In that study, X-ray ( the gold standard) had a PPA = 77% andNPA = 62% - just showing you that don’t need numbers greater than 70% to be auseful diagnostic test.
5- Thepatients condition may change over time and accordingly X-ray signs may changealso – thus it is not fair to compare a diagnostic test on day 1 ofpresentation with an X-ray taken on day 3 – and it sounds like in the USA in some casesmultiple chest Xrays were taken on the one patient – looking for that bit ofconsolidation on X-ray to confirm the diagnosis - the Americans are much more litigation focused. If the panel of doctors are diagnosing pneumoniabased on changes on X-rays on day 2 or 3 then that may not be a reasonablereflection of the patient’s condition on day 1. In my opinion – only theX-rays taken close to the time of initial assessment in ED should beused - but I'm not sure this occurred. In the Aus studies I'm sure only one X-ray would have been taken on the day of admission in almost all cases as X-ray does involve radiation that is harmful.
6- The most important data yet to come will be comparing ResApp diagnosis with the doctor in ED – what did thedoctor conclude after taking a history and examining the patient for 30 mins? –What was their initial diagnosis – before taking X-rays , blood tests , oxygensaturation readings?. This is what the ResApp test should be compared with. Is coughing into a smart phone as good as a doctor listening to the childs chest? if so - we have a winner.
7 - Considering the panel of 2 doctors could only agree in 67%of cases – I'd like to know what was the accuracy of ResApp for those 67% of cases? If 2 experts in Respiratory medicine cant agree on the diagnosis when looking at all the clinical notes including notes from multiple doctors examining the child, multiple blood tests taken, multiple X-rays done, what hope does any test have in achieving accuracy greater that 67%? Fromreading several studies it appears most of the studies classify Pneumonia into(a) Definite, (b) probable or possible and (c) not present. In the real world this is the important bit - all cases of probable and possible pneumonia are treated as well as the definite pneumonias. The definite ones are those with definite signs on X-ray. See below on X-ray reliability:
H.Melbye and K. Dale. Interobserver variability in the radiographic diagnosis ofadult outpatient pneumonia.
Acta Radiol. 33 (1):79-81, 1992. ( on pooragreement between clinicians on diagnosis)
Acute chest radiographs were obtained from 319adult patients with acute respiratory infections. Where a lower respiratoryinfection was diagnosed, follow-up chest radiographs were obtained in mostpatients. A radiologic panel diagnosed pneumonia in 21 patients. The agreementsbetween the panel and 3 independent interpreters, 2 residents in radiology, andone senior chest physician, were assessed. Also the reports given by thespecialist in radiology at the Department of Radiology were compared with the panel'sevaluation. While the kappa-agreements between the panel's interpretations andthose by the Department of Radiology and the consultant in chest medicine was0.71 and 0.72, respectively, the corresponding kappa-values between theresidents and the panel was only 0.50. The proportion of agreement whenpneumonia was diagnosed was 0.56 between the panel and the Department ofRadiology,
and 0.59 between the panel and the chestconsultant, compared to 0.36 between the panel and the residents.The study demonstrates the difficulty of diagnosing outpatient pneumonia andthe importance of experience. That was the abstract in 1992 - we see nothing much has changed!!
I’m not surewhat criteria were used in the US – but how about looking at the reliability of Definitepneumonia – ie no doubt = no disagreement between the panel in those 67% of cases, definite opacitieson X-ray on day 1 when the cough test was done – how did ResApp fare in that situation? – that is probably mostimportant – there will be a lot more sifting of this info - and maybe - maybe something might be worth submitting for the FDA.
OK - I hope I have clarified the info rather than confused you....
The facts are The ResApp technology is unique as a diagnostic screening test for most respiratory conditions. You cant expect cough sounds to be 100% accurate in diagnosing anything - but as a low cost / non invasive / non harmful screening test it has been proven to have great utility in Australian hospitals. It is soon to be trialed in a German hospital and I'd like to see an update of what is happening with Medicine Sans Frontiers in the 3rd world.
I might get around to discussing sleep apnoea one day which IMO has as much or more revenue potential as the cough diagnoses.
I won't be monitoring this thread too often - but happy to respond to sensible questions / comments. Look forward to the AGM next week.
