Bayesian Implications for Factitious Disorder

The next time you encounter a patient with an odd constellation of rare or poorly defined diseases (POTS and MCAS is a common duo), especially if a psychogenic one is on the list (e.g., PNES - not, I presume, intended to be sounded out, as it was proposed as a less offensive alternative to pseudoseizures), do yourself a favor and read one of Mark Feldman's excellent books on factitious disorder, such as the most recent, Dying to Be Ill.

These patients are wellsprings of iatrogenesis, but it is a special variety because the patients actively deceive physicians into unwittingly fostering complications through well-intentioned attempts to diagnose and treat disease. Several biases in medical decision making are exploited by patients with factitious disorder (and its close sibling, malingering, where the reward for the sick role is external - e.g., relief from work, military service or thwarting criminal justice proceedings - rather than internal), and I address some of them in this post.

First is what Feldman calls "truth bias" or a general tendency for people to believe what others tell them in everyday life, extended to medicine. As a result of this bias and other factors, Feldman estimates that less than 10-20% of factitious and related disorders are discovered and diagnosed. His book is rife with gripping anecdotes from actual patients who "came clean" from their deception and revealed the many ways they tricked physicians into believing they had serious disease. One talks about faking seizures and being frustrated that doctors did not believe her so she went outside and threw herself face first onto the concrete and bashed her nose a couple of extra times to assure that she was bloodied up convincingly. Another inhaled a cloud of baking flour and nearly died of asphyxia. Most tragic are the cases of factitious disorder by proxy, where babies and children are harmed and sometimes killed. I will state it clearly: when you are dealing with a patient with potential factitious disorder, the lengths they go to deceive are remarkable - stranger than fiction - the history is unreliable and you must consider the possibility that you are being lied to, as discomfiting as that is likely to be.

Compounding the cognitive dissonance experienced by the physician is another bias: physicians' tendency to (over)rely on "objective" testing. Consider a patient with factitious disorder who has "seizures" with atypical features, such as pelvic thrusting, speech during the episode, interruption of convulsions with pain. Inevitably an EEG will be ordered as an "objective" test to adjudicate whether true epilepsy is present. Suppose the EEG shows Interictal Epileptiform Discharges (IEDs, another unfortunate coinage by the neurologists, as this acronym has a baleful homonym): has a diagnosis of epilepsy been made? That depends upon the prior probability of epilepsy and the test performance characteristics. Let's run some numbers into the calculator on the sidebar. Lets use as an example the patient with pelvic thrusting whose "seizure" is interrupted by a painful stimulus, and give her a generous 1% probability of true epilepsy. (We should not even be entertaining the diagnosis with a probability this low, but it is an ineluctable course we are set on, and if we don't order it, somebody else will.) You will not be surprised that the sensitivity of an interictal EEG is low (25-50%), but you may be vexed to learn that the specificity is estimated to be 78-98%, far from perfect. If we use the optimistic values in those ranges (sens 50%, spec 98%) with our prior probability of 1%, we get a posterior probability of epilepsy of20%; if we use the pessimistic values (sens 25%, spec 78%) the posterior probability doesn't even change, as is always the case when sensitivity + specificity ~ to 100% (the so-called "rule of 100"). Yet, time and again this calculus will not be done, the positive test result will be taken as confirmation of epilepsy, the patient will be treated with AEDs (Anti-Epileptic Drugs, another homonymic acronym!) and the "seizures" will continue. Other drugs will be added and so on, ad nauseum, abetted by another bias: failure to see that the symptoms don't respond to the appropriate therapies and that this lack of response to treatment is itself disconfirming evidence of the provisional diagnosis. Instead of re-evaluating the validity of the diagnosis, the patient is plied with more and more drugs, increasing the risk of side effects. The result is often the comingling of malingering and true iatrogenic disease, clouding the picture even further.

This is just the tip of the iceberg. Suppose the EEG is negative. No worries, there will be another one and another one and another one, each with its chance of a false positive that will not be regarded as such. The possibility of syncope will probably arise because of collapse attacks, which will set in motion another cascade of testing with its own risk of false positives which likewise won't be recognized because the prior probability is ignored. Because these patients often have multiple complaints and high healthcare utilization in terms of emergency room and office visits, the volume of sundry testing will be very high with a concomitant rate of false positive results. Just like publication bias, the hits will receive more attention in the EMR than the misses.

A final mistake, if not a bias, is to fail to use Ockham's razor to try to make a unifying diagnosis. The unifying diagnosis is factitious disorder, the common cause of all the symptoms, and a far more cogent explanation than the constellation of Mast Cell Activation Syndrome (MCAS), Postural Orthostatic Tachycardia Syndrome, Seizures (POTS), Syncope, Apnea spells, etc, all in a young woman or man who at some prior juncture was perfectly healthy. Factitious disorder is not only a common cause diagnosis, but also a common sense diagnosis, and that's a major reason it is missed or made belatedly, after iatrogenesis been allowed run amok for too long.

These patients need empathy and compassion, but they also need a correct diagnosis, so they may be supported and treated without iatrogenesis. When somebody is Dying to be Ill, it is not compassionate to treat them to death.