VIRAL LOAD TEST ERROR MARGIN

[newbie76]

Hello there. I have read in various literature that the Viral Load test has a margin of error and here in the UK they seem to always make the example that if there's a shift between one test and another from 5000 to 15000 it doesn't necessarily mean that anything is wrong as a 3-fold increase is within the margin error of the test.

However an increase from 5000 to 25000 instead means that the VL has actually increased.

Up to here I am ok.

What I don't understand though is, does the same principle then apply for higher nubers then?? For example, if A VL of 50.000 becomes 150.000 is that still not worrying because it's a 3-fold increase and therefore withing the error margin or?

 

[newt]

Ok

maths..

You are right, the margin of error for a current viral load test is a factor of 3.  That means the result number indicates the middle of a range from 1/3rd the figure to 3 x the figure, for example:

Viral load result 50,000 ... Range 17,000-150,000

Okay, but ain't a viral load of 150,000 copies/mL massively more virus than 50,000?

No.

You can express viral load a different way, as a log value.  +/- half a log is considered (the beginning of) a significant change.

17,000 = log 4.2.
50,000 has a log value of 4.7.
150,000 = log 5.2 (rounding a bit). 

So both 17k and 150k are within half a log of 50k.

If you wanna convert your viral load into logs on Google. Just type "log(viral_load_number_here)" into the search box...eg "log(50)" returns 1.69 << undetectable 

A rough and simple way is to do viral load to logs is lose the 0 of the end of the viral load count. This is the same as a 1 log drop..eg log 150,000 = 5.2, log 15,000 = 4.2.

Newbie 76... hello ... in your case a change from 5,000 to 25,000 may be significant, but then it may not. It's a change of 0.7 log. Really you need at least 1 more viral load to judge. 5,000 indicates a viral load that could be as high as 15,000 and 25,000 one that could be as low as 8,500, so the margin of accuracy for these two tests overlaps.

Hope this helps

- matt

[megasept]

Ok

maths..
Hope this helps

- matt

Matt: I have never understood logs, now I think I might finally get it. Once again thanks for spreading useful knowledge here and everywhere throughout the land.

  -megasept

[newbie76]

Thank you newt... I don't understand everything you say, but I understand that a change from 50000 to 150.000 is within the error margins of the test and that answer my question!

By the way, my VL was 64 at the last count, not 25.000! That was just an example to see if I understood correctly!

Thank you!

[Moffie65]

Well Newt, I FINALLY GOT IT!!!!!!!!! 

Thanks so very much.

Love,

P.S. Yes guys, I've been + for 24 years and never understood it like I do now, so yes, you are always learning!

[newt]

I blush 

Handy links

Roche's HIV viral load slide rule with conversion widget, table << possible to analyse difference in 2 HIV viral loads

Janis & Friends hepatitis C viral loads with tables, terms explained and conversion factors for when comparing results from different tests

- matt

[tsw923]

Matt,

Thanks for the links and the explanation, but maybe I'm just too tired for it to sink in.  Please tell me if this is correct.  On my last test my VL was <75.  On the test before that my VL was <50.  My doctor told me not to worry as the 'sensitivity of the test' was not the same (I switchted health plans and therefore labs).  But back to the point, if I use google the log numbers are 1.70 and 1.88 respectively (rounding up).  So the difference is -.18.  And that's where I get lost.  Are we saying that it is statistically insignificant as it is not more than a 1/2 log difference?

Sorry -- I muddled and sleepy,

Ty

[milker]

tsw,

both tests show undetectable. You have 0 to 50 or 0 to 75 copies of the virus in the blood sample that was taken. You cannot correlate test sensitivity to your test results.

Milker.

[tsw923]

Milker,

Thanks for letting me know.  I guess I didn't know that there was different sensitivities to this particular test.  Is there any way I can know the name of the test that is being taken.  I'm getting over being embarrassed when I go in to give blood for the tests and more interested in what they are supposed to be taking.

Ty