Study Design for assessing New Clinical Sign where gold standard is not available


New Member
I am looking for some guidance as to what statistical tests I would need to do to know the significance of a study.

The situation is this:
Disease in Question is Sinusitis
Gold Standard for Diagnosis is CT, but is seldom done as not all cases require confirmation. Cost-effectiveness guidelines dictates using clinical criteria (2 Major or 1 Major & 2 Minor Criteria as defined) to determine if a patient is a case of Sinusitis or not.
Common clinical sign used is Tenderness to Pressure.
To know about the efficacy of a New Clinical Sign, a pilot study is performed to know if the new Clinical Sign is valid and to know the scope of evaluation needed to know its efficacy. 50 cases and controls are selected. Cases are selected based on the clinical criteria definition as above. Controls are healthy volunteers.
Cases and controls are gender matched

Tests for both Tenderness to Pressure and New Sign are performed on both controls and cases and data about number testing positive/negative for each test is obtained.
Now, to know about the efficacy of the New Sign what analyses are to be performed?

The sensitivity and specificity of Tenderness to Pressure are known. CT data is not available for the patients. The cases, as mentioned above, were defined based on the criteria, so are not gold standard proven.

Now how exactly is the pilot study result to be interpreted for significance?
McNemar's test was the one that was suggested, but I am not sure if it can be used as the cases are not determined by using the gold standard.
I would be really grateful if someone can point out which statistical test needs to be done to know the significance of the result.
As i understand it, Even for determining sensitivity and specificity, the case needs to be determined by gold standard. So, what are the parameters that can be defined here? Or can it be derived from the known parameters of the Tenderness to Pressure test? If yes, How?


Omega Contributor
None of the new criteria use any type of tenderness derivative? None of the patients had a head CT? Has this study already been conducted? How do you confirm sinusitis in patients without CT?