sensitivity and specificity in SPSS?

[ruhi]

hi
may it possible to calculate the sensitivity and specificity of a
procedure in SPSS especially the confidence intervals??? thanks

[Jeremy Miles]

You can use the ROC procedure to calculate sensitivity and specificity.
You can't get CIs though, through that, and I don't know another way
in SPSS.

Jeremy

--
Jeremy Miles
mailto:jn...@york.ac.uk http://www-users.york.ac.uk/~jnvm1/
Dept of Health Sciences (Area 4), University of York, York, YO10 5DD
Phone: 01904 321375 Mobile: 07941 228018 Fax 01904 321320

NOTE: New address from September 2006:
RAND Corporation, 1776 Main St, Santa Monica, CA, USA.
(New email and stuff too, but I don't know it yet).

[John Uebersax]

Hi Ruhi,

I believe that the DAG_Stat spreadsheet:

http://www.mhri.edu.au/biostats/DAG_Stat/index.htm

will calculate and report the confidence intervals for Se and Sp. You
just need to supply the four numbers of the relevant 2x2 summary table.

Hope this helps.
--
John Uebersax PhD

[John Uebersax]

Jeremy Miles wrote:

> NOTE: New address from September 2006:
> RAND Corporation, 1776 Main St, Santa Monica, CA, USA.

Hey Jeremy--fantastic! I worked there from 1987-1990. Have fun at the
beach :)

John

[Simon, Steve, PhD]

I'm working on a grant where I am the PI. All of my previous experience
was working on someone else's grant.

Anyway, I decided to adopt an informal tone in my grant. This has not
been received well by the first few people who have seen the grant.

For example, where I had written

> If you are a researcher, a member of an Institutional Review
> Boards (IRB), or a member of a Data and Safety Monitoring Board
> (DSMB), you need good statistical tools for the initial planning
> and the ongoing monitoring of clinical trials. In particular,
> you need to carefully consider the accrual rate--how rapidly
> are patients being recruited into your clinical trial.

a suggestion was made to write it as

> Members of Institutional Review Boards (IRB), or Data and
> Safety Monitoring Board (DSMB), need good statistical tools for
> the initial planning and the ongoing monitoring of clinical
> trials. In particular, they need to carefully consider the accrual
> rate--how rapidly are patients being recruited into the clinical
> trial.

In another place I write

> Consider a hypothetical research study that you started in
> January 1997 with the intention to recruit 12 patients per year (one
> per month) over a ten year period, for a total sample size of 120
> patients. By the end of June 2004, (roughly 7 1/2 years), the study
has
> enrolled 42 patients
>
> Clearly you have problems. Your actual accrual rate is a
> meager 5.6 patients per year, and now it is probably too late to fix
> things. You would have to recruit at a rate almost 30 patients per
> year over the remainder of the study to overcome this shortfall.
>
> Let's be honest--you knew it was a bit of a stretch to get 12
> patients per year, and now you have to more than double that accrual
> rate. Wouldn't it be nicer if you had noticed the problem two years
> into the study rather than 7 1/2 years?

and I am told that this is also a problem.

I am of two minds about this. Half of me says play it safe and adopt a
slightly more formal tone. The other half says that a breezy informal
style will make my grant stand out in a good way and get more attention.

I'll probably go with the play it safe mode, but I am curious what all
of you all think.

If you are curious about the grant, I have a weblog entry at

http://www.childrensmercy.org/stats/weblog2006/ColloquialTone.asp

that links to a PDF file of the current draft of the grant.

Steve Simon, ssi...@cmh.edu, Standard Disclaimer.
Look for my book "Statistical Evidence in Medical Trials"
newly published by OUP. For more details, see
http://www.childrens-mercy.org/stats/evidence.asp

[ruhi]

Jeremy Miles wrote:
> You can use the ROC procedure to calculate sensitivity and specificity.
> You can't get CIs though, through that, and I don't know another way
> in SPSS.
>
>
> Jeremy
>
>

> thanks for reply but i cant get Sp and CIs through the ROC procedure.?!

[Jeremy Miles]

ruhi wrote:
>
> Jeremy Miles wrote:
>> You can use the ROC procedure to calculate sensitivity and specificity.
>> You can't get CIs though, through that, and I don't know another way
>> in SPSS.
>>
>>
>> Jeremy
>>
>>
>> thanks for reply but i cant get Sp and CIs through the ROC procedure.?!
>>

You can get specificity, buy not confidence intervals. (Well, I imagine
that there's a way of programming SPSS to do this, but I'm not clever
enough to know how, and I've never seen anyone do it.)

J

[ruhi]

thanku, handy n excellent:)
Ruhi