sample size calculation for diagnostic accuracy study

[Shripada Rao]

Hi, we are planning to conduct a diagnostic accuracy study. How to calculate sample size for this type of study in Stata 16.0?

[Joseph Coveney]

For estimating sample size requirements, wouldn't you need to establish a few preliminaries? For example:

the objective of the study, which will vary depending upon where in the patient flow the diagnostic test is intended to be positioned (screening in low-prevalence population, confirmation in an enriched-prevalence screen-positive subpopulation) & assumption about prevalence generally,

what measure of diagnostic accuracy your study uses,

what the specifications are of your null- and alternative-hypothesis pair.

[Shripada Rao]

Thank you, Joseph.
The objective of the study is to find if the new test (index test) can predict the disease. It is a phase 1 diagnostic accuracy study.
The prevalence of the disease is 33%. The null hypothesis is that AUC for the index test is 0.5. The alternative hypothesis is AUC is 0.7. Regards

[Leonardo Guizzetti]

The following paper may be of interest to you:

Obuchowski, N. and McClish, D. 1997. 'Sample Size Determination for Diagnostic Accuracy Studies Involving Binormal ROC Curve Indices.' Statistics in Medicine, 16, pp.1529-1542.

[Joseph Coveney]

And if you're going to be using nonparametric method (you haven't said, despite the request in #2 above), then take a look at Sections V and VI of the original Hanley and McNeil paper. The paper that Leonardo helpfully points out will be useful if you use the binormal method.

James A. Hanley and Barbara J. McNeil, The meaning and use of the area under a receiver operating characteristic (ROC) curve. _Radiology_ 143:29-36, 1982.

[Shripada Rao]

Thank you Leonardo and Joseph, they are excellent articles and I found them highly informative. I also found this article (especially table 4) useful. They also provide information on specialized software for calculating sample sizes.

Nancy A Obuchowski, Michael L Lieber, Frank H Wians, Jr., ROC Curves in Clinical Chemistry: Uses, Misuses, and Possible Solutions, Clinical Chemistry, Volume 50, Issue 7, 1 July 2004, Pages 1118–1125,

https://academic.oup.com/clinchem/ar...7/1118/5640054

[Inaamul Haq]

There's a user-defined program Stata program, power_oneroc (and power_roc), for sample size calculation for diagnostic accuracy studies aiming to estimate AUC or compare AUCs of two tests.

[Carolina Hincapie]

Dear Inaamul Haq,

I just ssc install power_roc, but I cant not find the documentation for it use.
Do you know where I can find proper documentation for the commands?

Thanks in advance

[Inaamul Haq]

type

help power_oneroc

in Stata command window.

[Ariel Linden]

There is also a user-written package called "power_diagnostic" for you to consider

[Shripada Rao]

Dear Ariel and Inaamul, thank you for your replies. It's great to know that both packages (power_diagnostic and power_oneroc) have been written by Ariel. Amazing work! Thankyou