A National Academy of Sciences panel called yesterday for major changes in the outlines of the experiments the Veterans Administration has developed to study the health effects of the herbicide Agent Orange.

The protocols for the experiments are seriously flawed and need to be revised, the panel said.

For example, it said the questionnaire the VA intends to use when it interviews Vietnam veterans is filled with "ambiguous and esoteric terms" and places too much emphasis on diagnoses rather than identifying symptoms. The way the Defense Department plans to identify Vietnam veterans who should be tested also was criticized.

In May, 1981, the VA awarded a $114,288 contract to the University of California at Los Angeles to design the study.

But the government rejected UCLA's first report last November, saying it was so poorly done that it could not even be "classified as a protocol."

By the time UCLA submitted an acceptable experimental protocol in March, the VA already was considering ways to expand its study. This prompted congressional charges that the agency was intentionally delaying.

Last month, the VA announced that it would ask the Centers for Disease Control in Atlanta to perform the study. Negotiations between the two agencies are now in the works.

Regardless of which organization finally does the study, the position of the NAS panel is expected to cause even more delays in the study, which Congress ordered in 1979.

The study eventually could cost the government millions of dollars in liability claims by veterans who contend they have suffered ill health effects because they came into contact with the herbicide.

If either the VA or CDC ignores the academy's findings and proceeds with the study, the credibility of their work could be questioned.

On the other hand, if they accept the academy's recommendations, the study will have to undergo even more revisions.