Six months after the Food and Drug Administration withheld an internal finding that antidepressant medications were associated with an increased risk of suicide among children, a second staff analysis has arrived at the same conclusion.

The agency has not publicly disclosed either report, despite growing pressure from critics and Congress. Agency officials say they do not plan to discuss the data until a scheduled meeting in September, which would come nine months after British authorities warned physicians not to prescribe Paxil, Zoloft, Celexa and similar drugs to depressed children, and more than a year after the first concerns emerged.

The new analysis has renewed the complaints of critics that the FDA is moving too slowly to address the concerns about suicide.

One leading expert who reviewed the two internal analyses said they had changed his thinking about the risks. "I didn't have access to the proprietary data" reflected in the FDA analyses, said Steven Hyman, former director of the National Institute of Mental Health, who initially thought British regulators had overreacted. Hyman reviewed the analyses at the request of The Washington Post.

"Now I would say from the data in front of me, with the exception of Prozac, there is concern about the risk-benefit ratio of antidepressants in children," Hyman said.

FDA officials have declined to release the new analysis, first reported by the Wall Street Journal, and said they were still reviewing the information. "It would be premature to discuss any conclusions and results," said spokeswoman Kathleen Quinn. "Other people are going to have an opinion about it, but we are still reviewing it."

The new analysis comes amid growing concerns that crucial information about the safety and effectiveness of the drugs has been withheld from public scrutiny. Two-thirds of the trials conducted by drug manufacturers found that the medications performed no better than sugar pills, but details of the negative trials were kept from doctors and parents.

Joseph Glenmullen, a Harvard psychiatrist and author of "Prozac Backlash," a book about the side effects of antidepressants, said one company, Wyeth, had itself warned doctors not to prescribe its antidepressant Effexor for children: "The companies are coming to this conclusion, and the British came to this conclusion, and the Canadians and the rest of Europe and the FDA's own reviewer have come to this conclusion; why have they dragged their feet for over a year?"

The new "interim review" by FDA medical reviewer Tarek A. Hammad is a compilation of data with no narrative or interpretation. But both Hyman and Glenmullen said it showed an increased risk for children taking the drugs similar to that found in February by another FDA scientist, Andrew Mosholder.

Hammad's review found that compared with depressed children who got placebos, children who got antidepressants had 1.78 times the risk of making a suicide attempt or "making preparatory actions towards imminent suicidal behavior."

Mosholder found that children getting antidepressants had 1.9 times the risk of "serious suicide-related events."

Multiple congressional investigations are underway into the controversy. In a statement yesterday, the chairman of the Senate Finance Committee, Charles E. Grassley (R-Iowa) said: "It's been almost nine months since British regulators issued new recommendations, and it's been six months since Dr. Mosholder made his determinations. Now, given this new information, it's fair to ask if the Food and Drug Administration is taking too much time to draw a conclusion."

In April, the FDA's associate director for medical policy, Robert Temple, said in an interview that officials believed that a second analysis might provide a different answer than the one reached by Mosholder: "Andy thinks the results are unlikely to change from this analysis, and we are not so sure of that."

Temple said the agency decided to withhold Mosholder's findings because "we didn't think it was time to present a conclusion about a study because we as an agency didn't want to present it . . . in the absence of all the data; you want to be careful about reaching premature conclusions."

Agency officials said at the time that they were concerned about whether the companies had classified suicidal cases properly, and referred the question to Columbia University scientists. But in a memo dated Feb. 18, Mosholder warned the new analysis would waste time: "In my view, it is unlikely that the new information will alter the basic finding of an association of . . . serious suicide-related events with active treatment."

Given the risks, he urged the agency to take the interim step of officially discouraging the use of antidepressants other than Prozac for children. The FDA declined to do so but called for stronger warning labels to remind doctors to be vigilant about suicide -- for more than a decade, the agency has said that depression, not antidepressants, causes suicidal behavior.

"The whole thing is plain worrisome," said Hyman, who is currently provost of Harvard University. He said his reading of the two internal FDA documents suggested clinicians ought to use Prozac as first-line treatment for children's depression. If it failed to help, Hyman said, he would still cautiously use other medications, because the alternative -- leaving depression untreated -- could itself lead to suicidal behavior.

Jane Garland, a professor of psychiatry at the University of British Columbia, said she still thinks British regulators went too far. But after reviewing the two FDA analyses, she said the risk of suicide could not be dismissed: "I think the [FDA] assumption was if they looked at it more carefully it may disappear, and it hasn't disappeared." Garland recommended better tracking of side effects in clinical practice.

Glenmullen pointed out that no antidepressant apart from Prozac had demonstrated superiority to placebos in treating children's depression. If final FDA reviews confirmed the safety problems that Hammad and Mosholder found, he said, the lack of proven efficacy ought to discourage use of several drugs.

Both the FDA staff analyses calculated the risk of suicidal behavior for individual drugs as well as for all the drugs combined. Both were fraught with difficulties because the different studies used different measures of suicidal behavior.

Hammad and Mosholder found the risk of suicidal behavior varied widely among the drugs and between different trials of the same drug.

Hyman said more government-sponsored research is needed to objectively evaluate the issue. All the company-sponsored trials, for instance, excluded children who were suicidal to begin with. Including such patients might have showed the medicines lowered the risk of suicide, Hyman said. But just as easily, they might also have heightened concerns about the impact of the drugs on suicidal children.

"Maybe they are at the highest risk for these adverse events," Hyman said. "We just don't know."