lazy man's research

[Richard Goldstein]

I have some info culled from international registries of non-US rehab
medicine clinical trials

one of the questions of interest is whether a trial included subjects
over a certain age (e.g., 65 or 85); note that I do NOT have info on
the inclusion/exclusion criteria

for most trials I have the total N, but not for all trials

for some trials I have the mean and SD (though for some I only have the
mean) or I have info that can be used to estimate these values (I am
using the formulae suggested in Wan, X, et al. (2014), "Estimating the
sample mean and standard deviation from the sample size, median, range
and/or interquartile range", BMC Medical Research Methodology, 14: 135)

If I have the N, the mean and the SD AND if I can assume that the ages
are approximately normally distributed, then answering the over65 or
over85 question is easy. However, I am uncomfortable making this
assumption and I am looking for citations that provide guidance for
either (1) different distributions (esp skewed ones) and/or (2)
truncated distributions but where the truncation point is unknown. If I
could find such cites, I could do a sensitivity analysis re: the assumed
normal distribution answer

So, does any one know of any such citations or have other suggestions?

By the way, I know that I could do a bunch of simulations to get there
but I think this would be more expensive then my clients want to go.

I have received a suggestion that some help re: the normality assumption
might be available via national pop data and I will be looking into this

Best,
Rich

[Diana Kornbrot]

Age is more likely to be uniform than normal
It is worth classifying age rages and then performing ordinal regression
Classification also had advantage that one can choose ranges; young adult/student range, young worker, older worker, past retirement, very old

> --
> --
> To post a new thread to MedStats, send email to MedS...@googlegroups.com .
> MedStats' home page is http://groups.google.com/group/MedStats .
> Rules: http://groups.google.com/group/MedStats/web/medstats-rules
>
> ---
> You received this message because you are subscribed to the Google Groups "MedStats" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to medstats+u...@googlegroups.com.
> To view this discussion on the web, visit https://groups.google.com/d/msgid/medstats/fe206bc9-c469-3883-ab08-9cde224fbec7%40ix.netcom.com.

[Richard Goldstein]

sorry for not being clearer - I am only interested in the inclusion of
older people - at least at these ages, the distribution is not close to
being uniform (nor, at least in the few countries I am familiar with, it
is normal) - further, age is not a response variable in any sense; the
question is simply whether, based on only summary stats, I can infer
that people at least 85 (or, at least 65) were included in the trial

[John Whittington]

At 17:51 30/01/2022, Richard Goldstein wrote:
>sorry for not being clearer - I am only interested in the inclusion
>of older people - at least at these ages, the distribution is not
>close to being uniform (nor, at least in the few countries I am
>familiar with, it is normal) - further, age is not a response
>variable in any sense; the question is simply whether, based on only
>summary stats, I can infer that people at least 85 (or, at least 65)
>were included in the trial

Maybe I'm misunderstanding, but if the only information you have on
the age distribution in (some or all) of the trials is the mean and
SD, then you can only determine how likely it is that people >65
or >85 were included by making an assumption about the distribution
of ages to which the mean and SD relate.

To take a silly extreme small example, if the ages were (1, 2, 3, 4,
60, 61, 62, 63), the mean would be 32 and the SD about 31.6. If you
assumed a Normal distribution, you might then conclude that about 15%
of the population were over 65 and about 2.5% of the population was
over 95 - even though, in truth, none were over 65!

That is obviously a stupid extreme example, but far more modest
deviations from a Normal Distribution could lead to you drawing
appreciably incorrect conclusions about the proportion of people if
any!) >65 or >85 on the basis of just mean and SD.

Kind Regards,
John

>--
>--
>To post a new thread to MedStats, send email to MedS...@googlegroups.com .
>MedStats' home page is http://groups.google.com/group/MedStats .
>Rules: http://groups.google.com/group/MedStats/web/medstats-rules
>
>---
>You received this message because you are subscribed to the Google
>Groups "MedStats" group.
>To unsubscribe from this group and stop receiving emails from it,
>send an email to medstats+u...@googlegroups.com.
>To view this discussion on the web, visit

>https://groups.google.com/d/msgid/medstats/11d3c960-c062-5c73-0cc5-9ead4ce71e5e%40ix.netcom.com.

John

----------------------------------------------------------------
Dr John Whittington, Voice: +44 (0) 1296 730225
Mediscience Services Fax: +44 (0) 1296 738893
Twyford Manor, Twyford, E-mail: Joh...@mediscience.co.uk
Buckingham MK18 4EL, UK
----------------------------------------------------------------