It might be worth reviewing the NCCLS proposed guideline on "How to define, determine, and utilize Reference Intervals in the Clinical laboratory" document C28-P vol 12 No. 2, March 1992.
Mario D'Costa Ph.D. FCACB
Director of Clinical Chemistry and Core Lab
St. Joseph's Health Centre
30 The Queensway, Toronto
On, Canada, M6R 1B5
Tel: 416-530-6263
Fax: 416-530-6559
email: [log in to unmask]
>>> <[log in to unmask]> 09/12/01 04:03AM >>>
I agree with Mr Ohman's view, six data points outside the reference limit is
a good target. But with enzymes like ALT all the 'abnormals' will lie
above the limit. Low ALT activity is of no clinical significance.
Therefore, using nonparametric statistics, it is reasonable to set the lower
limit of the reference limit at zero, thus 5% above the reference limit may
be excluded to give 95% confidence, ie Mr Ohman's original figure of 120
would be acceptable.
This could NOT be applied to all enzymes of course, alkaline phosphatase for
example.
Howard Worth
King's Mill Centre, Nottinghamshire
> -----Original Message-----
> From: Sten Öhman [SMTP:[log in to unmask]]
> Sent: 12 September 2001 06:56
> To: [log in to unmask]
> Subject: Re: ALT Reference Intervals
>
> At 2001-09-11 16:32 +0100, Eric S. Kilpatrick wrote:
> >With the help of our academic colleagues who have taken blood from
> >individuals chosen randomly from GP lists, we have been able to better
> >establish local reference intervals for several of our analytes. One
> >result which has caused us concern is ALT where our quoted upper value
> >(97.5th centile) is likely to change from 36 to 52 iu/L on our Beckman
> >LX20s which read on the top side for this analyte anyway.
>
> You did not mention the number of subjects in you reference group. Unless
> the number is very large your reference limit may be a matter of random.
>
> If you studied 120 healthy subjects only 3 will have values above the
> reference limit. There is a good rule in statistics never to rely on fewer
>
> observations than 6. Therefore a reference range should never be
> established using a reference group less than 240 subjects!
>
> Mr Sten Öhman, PhD
|