The study raised tough questions: How soon is too soon to give adolescents powerful antipsychotic drugs? How about when they have not developed a mental illness, but seem headed down that path? Would the hope of averting a devastating psychotic breakdown outweigh the risk of side effects from possibly needless drug treatment?

The study, headed by Yale's Dr. Thomas McGlashan and funded by the National Institute of Mental Health and the pharmaceutical giant Eli Lilly, was published earlier this month in the American Journal of Psychiatry. Researchers at Yale and three other clinics recruited "help-seeking" patients, mostly adolescents, who hadn't developed schizophrenia but appeared to be on that path. The subjects, aged 12 to 36, responded to ads or were referred by clinicians.

Half the patients were treated with Eli Lilly's antipsychotic drug Zyprexa Olanzapine. The other half was given a placebo. Investigators asked: Can early drug treatment prevent psychosis?

Unfortunately, low recruitment numbers and a high dropout rate â€" 60 entered the study, and over two-thirds dropped out â€" left so few patients that researchers could not reach a definitive answer. The only conclusion they drew was that the drug might be able to delay psychosis, but likely can't prevent it.

The study has drawn heat from patient watchdogs and bioethicists who question if kids were exposed needlessly to the drugs. Zyprexa can have serious side effects â€" weight gain, diabetes, and seizures. Also, a Boston Globe investigative piece reported that the suicide rate in the clinical trials of Zyprexa and two similar drugs was "two to five times higher than the norm."

The drug's most prevalent side effect is weight gain: Some subjects gained as much as 40 pounds. McGlashan, in an email, said that came as a surprise: "When we started planning this study in 1996 Zyprexa was relatively new on the market. Weight gain was a known side effect but there was not yet a sense of the enormity of the weight gain."

The drug's dangers prompted patient advocate Vera Sharav of the Alliance for Human Research Protection (pictured) to complain to government officials while the study was under way. "In this experiment, healthy children â€" who are not capable of giving voluntary, informed consent â€" are being put at high risks of harm for experimental purposes," she wrote in a letter to the federal Office of Human Research Protections in 2000.

The OHRP responded by tightening informed consent procedures, rebuking investigators for consent forms that had been "misplaced or not signed," and asking them to more fully describe the risks of Zyprexa. But they did not find other aspects of the study unethical.

Dr. Sorell Schwartz, a pharmacologist and bioethics professor at Georgetown University, takes issue with the study's predictive tools: Can researchers really predict who will become schizophrenic? If not, aren't they unnecessarily exposing kids to harm?

To select subjects for the study, McGlashan used a diagnostic questionnaire to identify who had "prodromal symptoms" of schizophrenia, i.e. symptoms that indicated the youth would later develop psychosis. By McGlashan's own account, the method was unreliable, since doctors only know in hindsight whether a patient was psychosis-bound or not.

"The question is, can you really tell if a patient is prodromal?" asked Schwartz in a phone interview. "If you can't, there's an ethical question about giving these people drugs" because of the side effects. "I mean, these drugs are not chocolate ice cream."

In a recent interview with The New York Times, McGlashan echoed Schwartz' doubts and displayed a change of heart.

"I'm more pessimistic about all this now," McGlashan told the Times reporter. "I don't think the drugs can prevent full-blown psychosis, only delay it." He added, "I think more than ever we need to follow a group of prodromal adolescents who get no drug treatment to see more clearly what happens and refine our understanding of what the prodrome is."

Should this line of approach â€" early treatment on children â€" be further pursued? McGlashan doesn't seem so keen anymore. He told the Times he was approached to partake in a debate on early treatment of high-risk adolescents. He declined an offer to speak on the pro side, responding, "absolutely not."

However, in a subsequent email interview, he seemed to have shaken off these doubts. McGlashan said he tested the diagnostic questionnaire before the study and found that of the first 12 people who took it, over half became psychotic within a year. The method is still being tested for predictive accuracy, but "we regard it as accurate enough now to justify treatment research," he wrote.

Is early treatment worth the risks? "Aside from the weight gain" â€" yes, he wrote. Especially "for those who became psychotic, because being in the study meant that they could get standard treatment for psychosis as soon as they met criteria for psychosis.

"This usually meant they got treatment months before they probably would have gotten it had they not been in our program (it is know[n], for example, that the average time it takes someone with first psychosis to get into treatment is about 1 year). This is a major public health problem, and our strategy of early detection addressed this problem head on."

"I am very supportive of the idea of extending this line of research inquiry," he wrote â€" perhaps with a different drug. "An antipsychotic drug that does not induce such a degree of weight gain would be a natural next study."

Vera Sharav, the patient advocate, hopes lingering doubts will put an end to pharmaceutical companies attempting "unethical market expansion" to undiagnosed patients.

Others are holding out hope that early treatment may yet forestall a devastating illness that can damage someone for life. Howard Zonana, professor of psychiatry at Yale and longtime chair of the Yale-New Haven bioethics committee, acknowledged "I don't think we're pros at" predicting psychosis. But "If there's a way to delay the onset, that's a major advancement in the field."