ORLANDO – The mortality risk associated with prescription of antipsychotics or valproic acid for behavioral disturbances in patients with dementia is considerably greater than previously estimated, according to a massive national study.

This retrospective study included 45,669 matched pairs of Veterans Affairs patients over age 65 with dementia diagnosed during 1998-2009. Thus, there were 45,669 VA patients over age 65 with dementia diagnosed during 1998-2009 who were on one of the medications under scrutiny: haloperidol, olanzapine, quetiapine, risperidone, valproic acid, or a nontricyclic antidepressant. Each of these patients was matched to an older VA patient with dementia who was not on any of those medications and had not been for at least the past 6 months.

Investigators defined the number needed to harm (NNH) as the number of dementia patients who would need to be on one of the medications for 180 days in order to result in one additional death compared with nonuser matched controls. The NNH ranged from a worst-case scenario of 15 for haloperidol to 50 for quetiapine, Dr. Donovan T. Maust reported at the annual meeting of the American Association for Geriatric Psychiatry.

The patient pairs were matched based on demographics; comorbid medical and psychiatric diagnoses, including delirium within the prior 12 months; and a history of psychiatric hospitalization, noted Dr. Maust of the section on geriatric psychiatry at the University of Michigan, Ann Arbor.

The mortality NNHs found in this large study are considerably lower – that is, more unfavorable than in the two earlier meta-analyses conducted by other investigators. In the first meta-analysis, the estimated NNH for second-generation antipsychotics versus placebo was 100 (JAMA 2005;294:1934-43). More recently, a 2011 Agency for Healthcare Research Quality comparative effectiveness review found an NNH of 87 for second-generation antipsychotics when used in treating behavioral disturbances in elderly patients with dementia. The explanation for the larger, more favorable NNHs found in these meta-analyses is that they relied mainly upon studies of 8-12 weeks’ duration and so failed to capture the additional drug-related deaths that occurred over the course of the 180-day study in VA patients, the psychiatrist said.

In addition to providing a truer picture of the mortality risks of antipsychotics and valproic acid, the other advantage of the VA study is that it delineates the mortality risks for individual medications compared with no use at all. Earlier studies lumped antipsychotics together as a class, he noted.

"You can see that the risk is basically doubled with risperidone compared to quetiapine. So clinically I would use quetiapine as my first-line agent if I had to use anything. And I’ve stopped using haloperidol altogether," Dr. Maust said in an interview.

After the Food and Drug Administration issued its 2005 black box warning about the increased mortality risk of atypical antipsychotics in elderly patients with dementia, prescription of those drugs for the treatment of agitation and other behavioral symptoms in dementia outpatients dropped off, while the use of conventional antipsychotics climbed. With the 2008 black box warning regarding conventional antipsychotics, outpatient prescriptions for those agents also tailed off, but the use of valproic acid increased. However, the NNH of 29 for valproic acid seen in the new study shows this drug is problematic, too.

"It’s a bit like Whac-a-Mole where once one drug goes off the list, then people try something else. They’re such frustrating behaviors, and it’s such a distressing situation for the patient, the family, and the providers that it makes sense that people are trying alternatives," Dr. Maust commented.

The NNH of 158 for patients on antidepressant therapy seen in this study is probably not clinically meaningful in that it approaches the background mortality risk in this elderly, sick population, he added.

While the use of antipsychotics to treat behavioral disturbances in patients with dementia may have dropped off in outpatient settings since the black box warnings, that’s not the case in the inpatient setting – at least, not in the geriatric psychiatry inpatient unit at the Mayo Clinic in Rochester, Minn., Dr. Jung In "Kristin" Lee reported in a separate presentation at the meeting.

She presented a retrospective study of all patients hospitalized in the unit with a discharge diagnosis of dementia over the course of two study years: 2002 and 2012. In 2002, well before the black box warnings, 58% of 52 dementia patients were discharged on an antipsychotic agent, with quetiapine being the most commonly prescribed. Similarly, 65% of the 43 dementia patients discharged in 2012 were on an antipsychotic, with quetiapine once again being No. 1. In both years only a single dementia patient was discharged on a conventional antipsychotic, according to Dr. Lee of the Mayo Clinic.

Dr. Maust’s study was sponsored by the National Institute of Mental Health and the University of Michigan Program for Positive Aging. He reported having no financial conflicts, as did Dr. Lee, whose study was unfunded.

bjancin@frontlinemedcom.com

Antipsychotics’ risk in dementia revised upward