It is a good point elenacook about
Garbage in, garbage out. I also think that the study methods are flawed. I'm presenting here an attempt at analysis of the situation.
- assume good testing performed for ms/als/alz
- assume bad testing or no testing performed for lyme (with false negatives > 50%, false positives negligible)
- where ’lyme’ refers to any strain of Borrellia (including 30 strains distributed worldwide)
- analytical (not experimental) data includes counts of deaths resulting from either (ms/als/alzheimers) or lyme
- data is grouped by geographic region
- reliable data is not available to determine levels of lyme which are endemic to given area
- the following chart describes by row, ground truth variables for whether the lyme infestation is relatively high or low, as well as whether doctors who are making the diagnosis (dx) are capable of evaluating factors and trends (or ordering more accurate tests) to reliably determine the presence of lyme
- columns describe diagnosis at time of death
- values are keyed due to lack of space, and describe the implication of all of the factors described above and by position in the chart. IE, do all these factors make it more or less likely that lyme was present in the dead individual?
(Death case diagnosis) X (environmental factors) chart:
lyme dx neuro dx both dx
lyme awareness low 3 0 3
lyme awareness high 2 1 2
low endemic lyme 2 1 2
high endemic lyme 3 2b 3
Key:
0 : no information (is lyme endemic?)
1 : lyme not suspect
2 : lyme suspect
2b: lyme suspect (are doctors lyme aware?)
3 : lyme highly suspect
according to my analysis, in the lyme dx category in all cases lyme is likely. This is so even when another disease process is diagnosed.
But, in the neuro category, lyme is not likely when lyme awareness is high or when we know that there is low prevalence of lyme in an area. However, lyme awareness is poor in general, and we have no accurate information on distribution of lyme bacterias especially the strains that are not testable yet. The latter also assumes that the person never leaves their geographic area. So, these two rows are not good basis for publishing results.
We also cannot know that lyme is highly endemic, either. This removes the fourth row. Then, we are left with the first row - we know that we don't have good tools to analyze the prevalence of lyme. Therefore, we see that when a lyme dx is present, it is more likely to be accurate, but often this diagnosis is thrown out well before death unless a case of heart block, encephalitis, etc occurs acutely causing death. When a neuro dx is presented, it tends to stick around for the person's remaining years and can be attributed to death at any point inbetween. However, lyme was never suspected or checked with rigor - so how can we know that this person does not belong to the 'both dx' category? And in that case, what is the cause of death?
All of these things to consider, and still we are not even close to being able to prove or disprove a correlation or causation. Especially given that people travel so much, it is dubious that any such conclusion could be inferred from geographic associated data.
Post Edited (J Sandunes) : 11/17/2015 6:46:15 PM (GMT-7)