It is always an interesting phenomenon to see the headline grabbing news about how toxic psychiatric medications are killing people. At first it was just cult news, but these days it seems that some people can make a fairly good living at it. A knowledge of psychiatry or clinical experience is never a prerequisite. It always requires the reader to suspend their sense of reality and what they know happens in real life. That reality is that a family member, neighbor, or friend was having some very serious problems - saw a psychiatrist and got better. It also requires a suspension of belief in the tremendous history of what happens with untreated psychotic disorders both in terms of morbidity and mortality. Finally it requires suspension of a belief in the usual regulatory mechanisms. If so many people were dying from treatment - it would be obvious and somebody would be held accountable. Every state has medical boards that basically solicit complaints against physicians. Surely any group of physicians prescribing an inordinately toxic medication would come to light. You have to suspend all of these realities of course because none of it has occurred. Despite those reality factors there are any number of antipsychiatrists or people claiming to be critics who are basically using the same rhetoric warning people about the toxicities and how many people are killed by these medications each year. Some of their estimates are astronomical and suggest a clear and biased agenda. Actual community surveillance reveals an accurate picture and the medications with the highest complication rates are easily recognized by any psychiatrist or primary care physician.
That is not to say that the medications prescribed by psychiatrists are perfectly safe. As I just posted - no medical decision including one that involves taking a common medication is risk free. I spend a good deal - if not over half of my time warning people about side effects that will never happen, warning people about severe but rare side effects, managing side effects that do happen, and screening for potential side effects that might go unnoticed by the patient like electrocardiogram abnormalities or blood tests for a specific bodily systems. In 30 years of practice, I have diagnosed the most severe problems including serotonin syndrome, neuroleptic malignant syndrome, prolonged QTc interval, various degrees of heart block, arrhythmias, myocardial infarctions, strokes, drug-induced liver disease, agranulocytosis, diabetes mellitus, diabetes insipidus, hypo/hyperthyroidism and many other that were either caused by a medication or picked up as a result of my screening for a medical complication or pre-screening for safe use. But relative to primary care, the number of diagnoses in psychiatric practice for this reason is smaller. The most significant cause of mortality in psychiatric populations is cigarette smoke. The most significant number of medical conditions are pre-existing and if the psychiatric disorder is caused by an underlying medical condition - it is not common.
All of the factors in the first two paragraphs led me to read an article on the epidemiology of antipsychotics, antidepressants, and benzodiazepines in a well determined population and the effects on mortality in the June American Journal of Psychiatry. The authors have a number of studies that appear to use a similar epidemiological approach. For this study they identified cohort participants from national health care registers of all people 17-65 living in Sweden in 2005. They identified anyone receiving health care for schizophrenia or psychosis (by ICD-10 codes) and anyone on disability for schizophrenia. They also determined all of the antipsychotics, antidepressants and benzodiazepines dispensed from 2006-2010. They were classified by Anatomical Therapeutic Chemical Codes (ATC codes). They calculated cumulative exposures using the WHO defined daily dose (DDD) methodology. The WHO web site has a search engine that will let you search for the defined daily dose of medications. Examples for antipsychotic medications would include 10 mg for olanzapine and 5 mg for risperidone. The researchers summed the follow up days minus any hospitalization days and divided this into the sum of the dispensed medication. That allowed the subjects with schizophrenia to be broken into four DDD groups: 1) no antipsychotics, antidepressants or benzodiazepines during the follow-up, 2) low dose - small or occasional medications (0-0.5 DDD/day), 3) moderate doses (0.5-1.5 DDD/day, inclusive), and 4) high doses (>1.5 DDD/day). Using the olanzapine example that would mean a dose range from 0 - >15 mg/day cross all 4 groups.
A total of 1,591/21,492 or 7.4% of the cohort died in follow-up. That was 4.8 times higher than a control group of age and gender matched patients. The commonest causes of death were cardiovascular disease, neoplasms, respiratory diseases, and suicide in that order. No interactions were noted at the level of demographic variables. Mortality rates and hazard ratios for antipsychotic, antidepressant, and benzodiazepine use were calculated and the following observations were noted:
1. Any exposure to antipsychotics or antidepressants was associated with a lower rate of mortality (15-40% lower) compared to no use.
2. High exposure to benzodiazepines was associated with a 74% higher risk of death than no exposure. Benzodiazepine users had the highest mortality, highest risk of suicide, and more frequent visits to health care services.
3. In terms of cardiovascular mortality, only high dose antipsychotic use showed an equal mortality to no exposure to antipsychotics with low and moderate dose showing decreased mortality.
4. A sensitivity analysis of first episode patients showed that there was a decreased risk of mortality with exposure to low and moderate exposure to antipsychotics and increased mortality with exposure to moderate to high dose benzodiazepines. More striking is the fact that during the follow up period this was a cohort of 1,230 patients and 45 (4%) of them died. Most of the patients with first episode psychosis who I treated were otherwise healthy 20 year olds, illustrating the significance of this problem.
This is an excellent study from a number of perspectives. It looks at well defined data across a population that is generally possible only in Scandinavian populations. By contrast studies done in the US typically look at either incomplete retail pharmacy data designed originally for pharmaceutical sales or detailed health interview data that is based entirely on self report using long and detailed questionnaires. The study uses WHO methodology suggested for pharmacoepidemiological research. The follow-up period is during a times when most atypical antipsychotic medications are widely available. These are the drugs that are suggested as a source of higher cardiovascular mortality in psychiatric patients. The authors findings are discussed in light of several other studies that show similar effects.
The finding of this study will come as no surprise to acute care psychiatrists across the US. It is the reason why psychiatrists cover these settings despite the hardships involved. They know they are treating very difficult problems with very little assistance and that even in the absence of a continuum of care they can be successful. These psychiatrists are also aware of the medication toxicity and more importantly as this article points out - they can identify high risk patients and safely treat them. Despite the concerns about the metabolic effects of atypical antipsychotic medication there is an implication that other factors (like smoking) may be more significant in the development of cardiovascular disease (3). The risk of antidepressant and antipsychotic medication can be seen in an appropriate context in this study and that is lowering mortality rather than causing it.
The study also provides very useful guidance on benzodiazepine use. In my opinion, benzodiazepines should be used only briefly for the treatment of catatonia and acute agitation in patients with psychotic disorders. They should not be used on a long terms basis. I agree with the authors' idea that tolerance is a problem. When dose escalation fails or results in withdrawal and panic attacks or protracted insomnia, the risk for impulsive behavior and increasing depression is much greater. More frequent primary care visits can also occur due to tolerance and the need for dose escalation and more discussions of appropriate use. Treating this population in the United States is problematic because at a certain point, people can be safely detoxified from benzodiazepines only in an inpatient unit, and those services are widely unavailable. This study is a blueprint for quality assurance projects using the same methodology on electronic health records (EHR) across the country. Every clinical population should be examined using the authors' techniques and followed for outcomes and active interventions.
The reference provides an opportunity to see the realistic risk and benefits of treatment in people with high risk psychiatric illness. It also presents an opportunity to use this methodology to provide better treatment to people with the same illness and prescription profiles everywhere. Instead of using the EHR to catalogue useless full text information and track physicians, the authors methods can be used with much finer tracking of details like BMI, blood pressure, smoking status and other relevant lifestyle factors. Apart from the aspects of polypharmacy, the overall difference in mortality due to a diagnosis of a psychotic disorder needs to be addressed, and it needs the level of detail available in an EHR. Psychiatrists in major health plans using large databases could get active feedback in a very similar manner. The EHR could finally be used the way they advertised it a decade ago.
George Dawson, MD, DFAPA
3: Newcomer JW, Hennekens CH. Severe mental illness and risk of cardiovascular disease. JAMA. 2007 Oct 17;298(15):1794-6. PubMed PMID: 17940236.
