Antipsychotic Agents, Suddent Death, the New England Journal and Implications for Research and Disease Management

There have been a flurry of mainstream media reports (here and here and here, for example) about a study published in the New England Journal of Medicine (NEJM) that showed there was an increased risk of ‘sudden death’ among persons who use antipsychotic agent drugs.

The Disease Management Care Blog at your service. Unlike many in the media who quote from other’s press releases, it looked at the original study for numbers behind the media facts.

Briefly, the authors pulled 15 years’ worth of insurance claims data from Tennessee’s Medicaid program for persons with an average age of about 45 years who were using thioridazine (Mellaril is one brand name), haloperidol (Haldol is one brand name), clozapine (Clozaril is one brand name), quetiapine (Seroquel is one brand name), olanzapine (Zyprexa is one brand name) and respirdone (Resperdal is one brand name). These persons' sudden death rates were compared to matched persons with a similar insurance-medical profile who were not taking these drugs. That’s possible because every insurance claim indicates a diagnosis or a treatment and can be linked to age and gender.

What’s sudden death? Just what it sounds like. It’s a death that occurs within an hour of the onset of symptoms, though it’s usually thought of as simply keeling over. It’s usually but not always caused by a heart problem/coronary artery disease. It happens, but it’s rare, compared to other types of death, especially in this age group.

Antipsychotic agents block the action of dopamine in the brain and are widely prescribed for serious mental illness including psychosis. Because they also commonly have a ‘calming’ effect on users without causing sedation, they are also used by physicians for other mental illness when it is complicated by agitation such as Alzheimer’s Disease.

These drugs have been known for a long time to have side effects. They include weight gain (up to approximately ten lbs), the development of diabetes (it may go as high as 10%), higher blood pressure (more than 10 points systolic or the top number) and alterations (acquired long QT) in how electrical impulses are routed through the heart (which accounts for the electrocardiogram or ECG) in the course of normal beating (up to 3%).

The increase in weight, blood sugar and blood pressure could lead to an increased death rate from heart attacks. The alteration in the ECG could also lead to electrical instability in otherwise normal hearts. Whatever the cause, physicians have been long aware that patients who take antipsychotic agents have a higher risk of cardiac related death and/or sudden death. This risk appears to be particularly high when the newer antipsychotic agents (clozapine, quetiapine, olanzapine and respirdone) are used in elderly persons with Alzheimers Disease: the death rate can go from a baseline of 2.6% (remember, these are elderly persons) to 4.5% (in other words, the risk is increased by 1.9%) over three to four months. That's like a whopping 18% a year.

Based on the old information that there are heart problems associated with these drugs, is the information in the NEJM all that momentous? Decide for yourself: the authors found there were 895 deaths among ‘624,591 person-years’ (perhaps mathematically best thought of as 624,591 persons followed for one year) or a risk of 895 divided by 624,591 or 0.14 percent (just over a tenth of a percent) per year in the group of persons NOT taking antipsychotic drugs. Persons taking the drugs, on the other hand, experienced 223 deaths among 79,589 person-years or 0.29% per year. This overall death rate changed slightly depending on the types of drugs used, but the excess rate held up across all categories.

For comparison’s sake, check out the causes of death and their frequency in this population here. Therefore, while the risk of sudden seems very low for any single individual (3 in a thousand over a year), it’s a significant number from a population perspective (hundreds of Tennessee's citizens).

What can students of population-based health learn from this little gem of a study?

1. Once again, we’re witnessing the rise of rigorous insurance claims-based population studies that have an adequate comparison (control group) without having to rely on stodgy, time consuming expensive randomized prospective controlled clinical trials (RCTs). In fact, studies like this are more powerful than RCTs because they can draw on thousands of patients over many years and find important and statistically significant differences in the range of tenths of a percent. Disease management organizations also preside over huge data bases in their information systems and should draw a lesson from this study. They can and should also be doing this kind of research.

2. Disease management organizations that offer mental health programs have an additional reason to be vigilant about the use of antipsychotic agents in their patients. Since antipsychotic agents can mean the difference in being able to function for persons with chronic psychosis, the low risk of sudden death, when these drugs are truly needed, is outweighed by their benefit. On the other hand, use of these drugs in children or the elderly for their calming effect is dangerous (and promotion on that basis by pharmaceutical companies is illegal and expensive), especially when other approaches work just as well and are safer. Disease management organizations have a role to play (or perhaps a market opportunity?) in keeping that to a minimum through patient coaching and provider feedback.

Disease Management for Severe Mental Illness

We can telephonically engage/coach/monitor and help persons with diabetes, asthma, heart failure, COPD, coronary artery disease – but how about serious persistent mental illness? Check out this interesting paper published in the latest issue of the American Journal of Managed Care by Paul F. Cook, Suzie Emiliozzi, Corey Waters and Dana El Hajj of the University of Colorado, ScriptAssist and Centene Corporation (the latter two entities paid for the study).

This was a study involving outpatient Medicaid beneficiaries who were taking anti-psychotic medications for more than 30 days. 210 candidates were identified for nurse-based telephony, but only 59 (28%) could be reached on the phone. Of these, 8 declined to participate, leaving 51 intervention patients. The remaining 151 formed a comparison group. The results that caught the eye of the Disease Management Care Blog were that intervention patients visited ERs an average of about 1 time per year, versus 5 in the comparison group. While the comparison group initially had higher pharmacy compliance rates in the first month, by 6 months is was clear that the intervention group was doing much better at taking their anti-psychotic medications at 48% vs. 26%.

This was far from a perfect study. The intervention and control groups are probably not comparable, since patients who can be reached by telephone and agree to participate are probably more compliant in general and are more likely to avoid ERs and take their medicines as prescribed. That’s called selection bias. There was also insufficient detail about the protocols used by the nurses to help the patients.

Think of this as a pilot study. That being said:

Among a population of severely mentally ill outpatients who are notorious for not taking their antipsychotic medications, the DMCB learned that it’s possible to reach a quarter of them by telephone and engage them in regular follow-up. This may reduce ER use and it may increase medication compliance. This is worth more study.

There were other lessons learned from this paper. The authors pointed out that there was a delay between the time patients got their qualifying medications and the telephony was initiated. Another was the use of a masked “unknown” caller ID when patients were called (to protect confidentiality), which in turn prompted patients to not pick-up.

Despite all the limitations, the DMCB likes this study because it’s disease management outside the usual big 5 of asthma, diabetes, heart failure, COPD and coronary artery disease. It uses a well known approach to a population that is very difficult to manage with high levels of avoidable utilization and complications. It was performed in a community setting with high generalizability and had the wherewithal to not only implement a program but simultaneously study the impact at the same time.

It is common in research settings to conclude manuscripts like this with the adage that more research is necessary. The authors of this little gem did that also. However, the DCMB would go one step further and suggest (unless there are better ideas out there) that telephonic follow-up/coaching of persons with mental illness who are on antipsychotics be expanded and that more research be performed to better assess what works and what doesn’t.