Kevin, MD, recently wrote a piece on “defensive medicine”, which he defined as

the deviation from sound medical practice to avoid the threat of malpractice litigation…. This can range from “positive” defensive medicine, like ordering unnecessary tests, referring to consultants, or performing unneeded procedures; to “negative” defensive medicine, like avoiding high-risk patients or procedures.

I initially surmised that psychiatry is not as susceptible to litigation and thus, is not as prone to practicing defensive medicine.

Upon further reflection, however, this is untrue. There are many examples of defensive medicine in psychiatry; let’s consider the role of hospitalization in preventing suicide.

There is absolutely no evidence that hospitalizing people who are endorsing suicide actually prevents suicide, both in short- and long-term follow-up periods. No one will ever pursue this study because it is ethically unsound. To apply the “gold standard” of scientific evidence to this question, the study would have to gather a large group of people endorsing acute suicidality (what would this definition mean? suicidal ideation only? an actual attempt? acts of self-harm that are not intended to end life, such as cutting? some combination of the above? something else entirely?). Members of this population would be randomly assigned to either hospitalization or something else that is not hospitalization (sent home? follow up appointment the next day?). If the “something else that is not hospitalization” is a return to the community, that could possibly qualify as “placebo”, but the fact that someone presents to the mental health field with suicidality is an intervention in of itself. And there is no way to “blind” this study: everyone will know who is hospitalized and who is not.

Further complications of this question include the context of the suicidality: Should diagnoses be considered, or should the focus be on suicidality only, regardless of “cause”? (For example, the person who is intoxicated with alcohol and is endorsing suicide is not the same as the person who is psychotic and is endorsing suicide in an attempt to “beat the aliens to it”.) Is there a difference between the individual who is enrolled in mental health services and followed the instructions of his provider to seek psychiatric help, and the individual who has never interacted with the mental health system and was literally pulled off of the bridge by firemen? Do we take into account the person who has been hospitalized over twenty times for suicidal ideation or attempts, versus the person who has never thought about it at all in the past?

It’s complicated.

In fact, there is a paper that suggests that the highest risk of completed suicide occurs during the first week following admission to and the first week following discharge from the hospital. The data from the paper also suggests that many other factors affect the suicide risk (shorter hospitalization lengths increase risk; mood disorders affect the risk the greatest; women who have been admitted in the past have a higher risk; etc.).

However, while this paper provides information, we cannot draw many firm conclusions from it (i.e. hospitalization causes suicide). First of all, the data from the paper came from Denmark (the country maintains a national population registry)—can we apply Danish information to Americans? the British? Australians? Secondly, we don’t know what the suicide risk is for people who are not hospitalized; that is, although the rates for suicide around the time of psychiatric hospitalization are at their highest, would the rate be significantly higher if people weren’t hospitalized at all? We can estimate how many people kill themselves, but we don’t know for sure—the health care system only sees those people who (1) failed in their attempts (or were otherwise “caught in the act”—pulled off of the bridge, had the gun yanked out of their hands, etc.) and (2) admit to thinking about it and are presumably ambivalent about suicide.

We never hear from or see the people who successfully kill themselves.

How does this relate to defensive medicine? Some psychiatrists routinely hospitalize patients (through coercion, either with or without the “help” of friends, family, et al.) who endorse suicidality when “clinical suspicion is low” that the individual will actually commit the act. The belief is that if someone is in the hospital, the risk of suicide drops significantly—after all, the ward is locked, there is less access to means of killing oneself, and there are people (nursing staff) presumably monitoring the situation. There is also the sense that the psychiatrist, by admitting this person, is “doing something”, versus asking the patient to return for an outpatient follow-up appointment the next day, which seems more like “not doing much”. (This drive to “do something” is also what propels physicians to prescribe medications that are not necessarily indicated.)

Sometimes, hospitalizing patients is not the most therapeutic or appropriate approach (for example, for people who endorse chronic suicidality). However, to reduce risk and prevent litigation, the patient ends up in the hospital to prevent a potential malpractice lawsuit against the psychiatrist. Psychiatrists are human and humans cannot predict the future. Humans can, however, assess risk—and some people are more risk-averse than others.

(To be clear: Hospitalizing someone does not mitigate the suicide risk. Patients do kill themselves in the hospital.)

The lack of data in this issue, though, reminds me of the BMJ parachute paper:

As with many interventions intended to prevent ill health, the effectiveness of parachutes [to determine whether they are effective in preventing major trauma related to gravitational challenge] has not been subjected to rigorous evaluation by using randomised controlled trials. Advocates of evidence based medicine have criticised the adoption of interventions evaluated by using only observational data. We think that everyone might benefit if the most radical protagonists of evidence based medicine organised and participated in a double blind, randomised, placebo controlled, crossover trial of the parachute.

When jumping out of an airplane, one uses a parachute out of common sense. Can the same be said of hospitalizing someone who is endorsing suicidality? Does evidence-based practice only complicate (and obfuscate) straight-forward issues?

I don’t know the answers to these questions. I hope that there comes a day, though, where doctors and patients trust each other more… and that doctors can comfortably focus their energies on their patients instead of on themselves.