Lilly's Zyprexa
Acted Similarly
To Cheaper Drug

A study comparing a pennies-a-day schizophrenia drug with a far more expensive and widely prescribed Eli Lilly & Co. drug found no difference between the two in reducing schizophrenia symptoms and improving quality of life.

The results of the Lilly-sponsored study, published in this week's Journal of the American Medical Association, run counter to those of some earlier studies. In those earlier studies it was found that Lilly's Zyprexa -- the more expensive drug -- improved symptoms and quality of life and lowered health-care costs when compared with Haldol, largely by cutting down on hospitalization of schizophrenic patients.

Indeed, before they began enrolling patients, the authors of the JAMA study, a randomized trial that followed 309 military veterans for a year, believed the results would mirror the prior studies' findings and initially were surprised by their results. Upon further review, the principal author, Robert Rosenheck, said he found some problems in the design of some of the major prior studies.

"We need to rethink what we're getting for what we're paying," said Dr. Rosenheck, director of the U.S. Department of Veterans Affairs Northeast Program Evaluation Center in West Haven, Conn., and a professor of psychiatry and public health at Yale Medical School. "The profound issue in health care is do we pay anything -- regardless of the price -- for a statistically significant benefit?"

The participants in this study, which was conducted by doctors at 17 VA hospitals, don't mirror the general population: Nearly all were men -- with an average age of 46 -- who had been suffering from schizophrenia for about two decades. Still, in an era marked by sharply rising health-care costs, the study is likely to raise questions about what many doctors have seen as a costly but worthwhile first-line treatment for a debilitating disease. According to the study, about two million people in the U.S. suffer from schizophrenia, a psychotic disorder marked by delusions and hallucinations.

Zyprexa didn't reduce hospital stays, according to the study. Zyprexa was associated with substantially greater costs, ranging from $3,000 to $9,000 a patient annually. Zyprexa typically costs the Department of Veterans Affairs $8 a day per patient, while Haldol costs six cents a day, the agency said.

Alan Breier, Lilly's chief medical officer, said he didn't believe the study would prompt doctors to revert to prescribing Haldol or other older drugs before Zyprexa. Instead, he said, doctors need to look at the "totality of the literature" on Zyprexa. The drug had world-wide sales of $4.1 billion during the year ended Sept. 30 and makes up more than one-third of the Indianapolis drug maker's revenue. "I think it's important not to draw black-and-white conclusions from one study," Dr. Breier said.

Patients were randomly assigned to Zyprexa or Haldol and neither the doctor nor the patient knew which drug was being used. Patients taking Haldol also were given another drug to prevent tremors and other Parkinson's-like side effects of Haldol; that drug also costs pennies a day. (Zyprexa patients were given an inert companion pill to mimic the side-effect medication taken with Haldol.)

Patients taking Zyprexa were less likely to suffer from mild akathisia, a feeling of inner restless, and more likely to perform slightly better on tests that measure memory and fine movements. When adjustments were made to account for those who switched medications, the patients in the Zyprexa group also had fewer symptoms of tardive dyskinesia, a neurological problem that causes repetitive, involuntary movements, such as smacking lips, grimacing and rapidly moving arms and legs. But none of those benefits, when assessed by patients and trained raters, led to a greater quality of life for patients using widely accepted measures, the researchers concluded.

Still for some doctors, those benefits are enough to merit prescribing Zyprexa as a first-line treatment. "Even mild akathisia can be tormenting," said Stephen R. Marder, a professor at UCLA Neuropsychiatric Institute.

Zyprexa patients in the study also were more likely to report substantial weight gain than those on Haldol and the side-effect drug. That is a concern because the Food and Drug Administration recently required Lilly and other makers of drugs in that class to carry warning labels on the drugs about a possible risk of diabetes.

In the study, patients unhappy with their treatment could switch to another medication. The data were analyzed to make sure that medication switches didn't bias the results.

The study set out to recruit about 600 participants and ended up with about half that, although it had enough patients to draw statistically valid conclusions. Nonetheless, Dr. Breier, of Lilly, said the results could have been skewed by 28 patients for whom treatment costs exceeded $50,000. Of those "outliers," 17 were in the Zyprexa group, a Lilly spokeswoman said.

Dr. Rosenheck says Lilly suggested excluding those higher-cost patients from the analysis, which he refused to do. He crunched the data using four different methods to address the skewed distribution, each time coming up with the same results. "They began to suggest things that I did not feel comfortable publishing under my name," he said. "When we do research, we generally don't throw out data." A Lilly spokeswoman declined to comment.

Dr. Rosenheck knows of no prior studies comparing Zyprexa and Haldol that used a companion drug to prevent Haldol's side effects. Instead, he found some studies used a companion drug once the symptoms emerged. That is a problem, he said, because telltale symptoms let patients and doctors know that Haldol was used, prompting some patients to drop out. In addition, he said, some side effects may be mistaken for symptoms of schizophrenia. All of those factors could bias the results, he said.

Write to Patricia Callahan at patricia.callahan@wsj.com

Updated November 26, 2003