Photo of John Sullivan

We’ve never been deferential to differential diagnosis – at least, not in litigation.  There’s no need for a diagnostic tool when you’ve already got the diagnosis.  And plaintiffs almost always come to litigation with a diagnosis.  What’s generally at issue is causation, not the diagnosis.  Yet, more and more, doctors serving as experts for plaintiffs in litigation claim that they used a differential diagnosis to determine the cause of the plaintiff’s disease, and it’s always the drug or device.  These conclusions are sometimes quite litigation friendly – the product “was a substantial contributing factor” in the development of the disease or illness – a conclusion that sounds as non-medical as the process that arrived at it.

But today’s post – as much as it might want to be – isn’t about whether the use of a differential diagnosis is a scientifically reliable method for determining causation.  A number of courts have already accepted that it is, so we at times need, as a practical matter, approaches that undermine an expert’s application of the method.

Fortunately, as illustrated in In Re: Trasylol Prods. Liab. Litig., 2013 U.S. Dist. LEXIS 36012 (S.D. Fla. Mar. 13, 2013), there can be a lot to work with.  The plaintiff in Trasylol claimed that the use of the drug Trasylol during the decedent’s mitral valve surgery caused the decedent to suffer heart and renal failure a few months after the surgery and ultimately die.  Id. at *29.  Once in litigation, the plaintiff’s specific-causation expert, a nephrologist, used a differential diagnosis to determine that Trasylol was a “significant contributing factor” to the decedent’s death.

But the expert had two significant problems with his opinion, problems that we’ve seen in practice before: (1) he didn’t consider all the data (i.e., all the medical records and history) and (2) he didn’t reliably apply his differential diagnosis.

The first problem – not considering all the medical records – was a big one.  You know a deposition has started well when the plaintiff’s expert tells you that he didn’t review all the medical records.  Here, not only did he not review the decedent’s medical history, he didn’t review all of the records related to the decedent’s surgery, the event upon which the litigation was based.  This never looks good to a court:

[The plaintiff’s expert] made his conclusions with a glaringly incomplete medical history. Specifically, in making his conclusions, Dr. Blond never reviewed any records predating [the decedent’s] August 2006 hospitalization for her surgery, nor did he have a complete set of medical records for the August 2006 hospitalization.

Id. at 49-50.  This led to a number of embarrassing admissions by the expert.  He testified, for instance, that he did not see a history of smoking or past heart attacks, only to admit that the decedent did have such a history once he was shown the records at his deposition.  Id. at *49-51 n.16 & 17.  He also described a high creatinine reading taken before the decedent’s surgery as an anomaly because there was a subsequent low reading taken after the surgery, only to be forced to admit that the post-surgery reading was likely low because the decedent had post-surgery fluid retention.  The fluid retention was mentioned, once again, in medical records that the expert hadn’t reviewed.  Id. at *52-55.

The expert’s incomplete review of the medical records led to other methodological problems.  He missed, for instance, a whole host of risk factors that could have led to the decedent’s post-operative decline in kidney function:

[The plaintiff’s expert], in both his Report and his Deposition, failed to sufficiently consider [the decedent’s] diabetes, congestive heart failure, smoking, diuretic treatment, atrial fibrillation, hypertension, and contraction alkalosis at the “rule in” stage. Id. at 49.

It’s important for an expert applying a differential diagnosis to consider – or “rule in” – all possible causes of the illness and then systematically and scientifically eliminate – or “rule out” – those that could not have caused the illness.  Id. at *46-48.  But this expert couldn’t even get that analysis off the ground.  He didn’t know the plaintiff’s risk factors, so he couldn’t reliably consider them and then rule them out.  This meant, as the court found, that he couldn’t reliably claim that Trasylol caused or contributed to the decedent’s death:

[A]t no point during his deposition, does [the plaintiff’s expert] explain why, despite her moderate to high risk of renal failure, and despite the surgeon’s anticipation of renal insufficiency, her complex surgery, her previously unknown or partially known history of myocardial infarction, recreational drugs, smoking, and hypertension, [the decedent’s] “‘mild’ acute renal injury” was attributable to Trasylol.

Id. at *54.  The court excluded the expert’s opinion and, without a specific causation expert to support plaintiff’s claims, granted judgment on all claims for the defendant.

Now, this case was a bit of a slam dunk.  Frankly, it’s one of the worst – if not the worst – applications of the differential diagnosis method by an expert that we’ve seen.  And yet it contains good – if extreme – examples of weaknesses that can be exploited when a plaintiff’s expert tries to find specific causation by applying a differential diagnosis.  Even if the expert reads all or substantially all of the relevant records, there can still be problems.  These experts often have difficulty finding a scientifically or medically reliable method to rule out risk factors or, in the face of the patient’s medical history and longstanding risk factors, explain why or how the drug or device could suddenly become a substantial contributor to the disease or illness.  All of this is ripe for exploration at the expert’s deposition or even at a Daubert hearing.

So, yes, we don’t see the scientific or medical basis for determining causation via a differential diagnoses.  But since we see a lot of it, and see courts that regularly allow it, we’re glad to also see that there are meaningful ways to attack it.