org355: Mystery disease in Canada / pre-natal screening

1. What study design is most suitable for investigating a potentially new disease of unknown cause?

2. What is the outcome definition? What is the implication of having a ambiguous definition on the search for causes?

3. Would an RCT be required to confirm the cause?

And a bonus article from The New York Times - notable for some excellent examples of data visualisation, and for hitting on most of the main points raised in the 'screening' session.

1. How can a test be 99.9% "accurate" and yet have very poor positive predictive value?

2. Under what conditions should we routinely screen for extremely rare conditions?

Re: Mystery disease in Canada / pre-natal screening

by | FATHIMA MINISHA - Monday, 3 January 2022, 12:07 PM

Hi Sujit and everybody....
Happy New year to all!! The new year is beginning with COVID again.. I hope everybody is able to keep up their morale and find the motivation to work towards all their plans for the year...

1) We have the unknown disease- we have cases... and therefore we would be able to recruit healthy controls from the same population preferably age and sex matched...and look back at all possible exposures- genetic environmental etc... So here a case control (preferably matched) study would be one of the best designs

2) The outcome definition would depend on the cases noted till date and decided by a group of experts- the issue is that the definition should be sensitive enough to be able to detect most cases but also at the same time avoid too much misclassification. In this case probably a combination of most commonly noted symptoms and brain imaging would be required to define a case. Having an ambiguous definition would mean misclassification- non-differential and therefore the all effects would be pulled towards the null. This would greatly affect the quest to determine the exposures leading to the disease.

3) The answer would be "No". A well conducted case control could give an idea about the exposure. Based on the preliminary studies probably a cohort study can be done to further assess causality. However, an RCT would not be ethical here because the disease results in a neurological damage and it would be unethical to randomize healthy ppl to different possible causative exposures and wait for this disease to occur.

Another very interesting read about screening tests...
1) A test could be 99.9% accurate in the sense if there is actually a disease the test detects it... but that does not mean that all of those who test positive actually has the disease. Positive predictive value is the proportion of those who test positive that actually have the disease- and that can be really really low when a test has a high false-positive rate. It could detect 99.9% of the actual disease out there, but if the disease itself is very rare, the test will give out a lot of false positives and drastically reduce the positive predictive value...

2) Routine screening in extremely rare conditions should be restricted to high risk groups only- in whom the probability or risk of disease is much higher such that the test could have better PPVs.

Fathima

Re: Mystery disease in Canada / pre-natal screening

by | ZACHARY MOST - Tuesday, 4 January 2022, 2:37 PM

Hello Sujit and Fathima.

I agree with Fathima's answers with some additional comments

1) I agree that a case-control study would be best to determine the cause or risk factors for the disease. Even the question about how much toxin detected in post-mortem brain tissue is "normal" could be investigated with a case-control study (in this case controls would be age matched deceased individuals which could be quite difficult to recruit). But I would also add that for "investigating" a new disease, often the first step would be to publish a case series, which would be a descriptive study of the known cases and could help establish a case definition

2) The biggest problem with this article is that the outcome is defined quite vaguely. I really have no idea what the outcome definition was for this disease. I agree with Fathima's points on this. But would also add that a case definition that is too sensitive (and not specific enough) would also lead to a false elevation in the number of reported cases. For a rare disease this could mean that many (if not most) cases that meet the case definition are actually false positives.

3) I agree. In theory a randomized trial on an animal model could be considered though,

1) Agree, you always need to consider pre-test probability when interpreting tests results

2) I would add that you can consider screening for extremely rare conditions when: a test with good sensitivity and specificity is available, the test is cheap and easy to administer, and early diagnosis of patients with the disease would lead to meaningful differences in their outcome (i.e. early diagnosis leads to cure or prevention of severe outcomes). An example of this would be newborn screening for metabolic disorders like phenylketonuria (which we do in the USA).

Zach

Re: Mystery disease in Canada / pre-natal screening

by | Sujit Rathod - Tuesday, 4 January 2022, 3:42 PM

Welcome, Zach, and many thanks for responding. And to Fathima for getting the discussion off to an excellent start.

I hadn't thought of an RCT with an animal model. Good suggestion! Though an RCT (usually) assumes a relatively speedy manifestation of the disease after exposure. What if the exposure arose in the childhood of these individuals?