What Is the Role, if any, of Differential Clinical Diagnosis in Trials?
     Physicians often make treatment decisions based on conclusions that have been derived to be the nature and cause of the illness. This is achieved through the method of differential diagnosis that is commonly defined as "the determination of which of two or more diseases or conditions a patient is suffering from by systematically comparing and contrasting their clinical findings."⁽¹⁾ Using this method a physician brings together information suggesting that an agent can cause a specific disorder. Using information gathered from the patient's exams along with patient history and research on the disorder, the physician could draw conclusions about the cause and nature of the patient's injury or illness. The quality and quantity of information available for consideration varies, as does the certainty of the analysis.

     In some cases examination of the patient reveals clear symptoms of a "signature disease"⁽²⁾ known only to be caused by one agent, such as mesothelioma, known to be caused by asbestos. These diagnoses are an accepted method of proving specific causation in such a situation. When a signature disease is not present, clinicians use information from animal studies, chemical analysis, epidemiology, patient history, or the timing of symptoms to make treatment decisions. Logically, this process has been likened to the making of soup⁽³⁾ or assembling a "mosaic,"⁽⁴⁾ putting together information from a variety of sources using professional judgment. The degree of certainty to a diagnosis varies with the types and quality of information used in its making, and is perhaps at its lowest when the only link between alleged cause and illness is temporal.⁽⁵⁾ The suitability to the courtroom of this method so common in the exam room is a subject of debate.

Should a Clinician's Testimony Be Admitted As Evidence of Causation?

A. The Case For Admissibility
     Commentators supporting admission of differential diagnosis often argue that its long-standing and generally successful use by the medical profession to draw conclusions good enough to guide medical treatment demonstrates its legitimacy. While recognizing that it does not provide "absolute certainty,"⁽⁶⁾ it is praised as providing information based on "wisdom, probity, and approbation"⁽⁷⁾ from the "actual practice of medicine"⁽⁸⁾ based upon a mix of medical knowledge, intuition and judgment.⁽⁹⁾

     A number of courts have admitted testimony based on differential diagnosis after Daubert. The Fourth Circuit in Benedi v. McNeil upheld the admission of differential diagnosis linking illness to the combination of Tylenol and alcohol.⁽¹⁰⁾ The court determined that methodology good enough for the "day in and day out" practice of a diagnosing physician is sufficient for use in federal courts even if not supported by epidemiology.⁽¹¹⁾ A similar logic was used by the Second Circuit in McCullock v. H.B. Fuller,⁽¹²⁾ allowing physician testimony on causation based on differential diagnosis using patient history, treatment, and pathology, holding that faults in the physician's analysis went to the testimony's weight rather than its admissibility. McCullock was cited in Becker v. National Health Products to support the admission of two physicians' testimony on illness caused by a dietary supplement because of their professional experience, review of relevant literature and examination of the client.⁽¹³⁾

B. The Case Against Differential Diagnosis
     A number of courts have taken issue with this approach, attacking it for its inability to provide logical support for a position, or reliance on bad methodology in reaching its conclusions. Differential diagnosis may be appropriate to a health care situation where the most likely cause of an illness must be identified and acted upon to save lives but inappropriate as a method of fact-finding in adjudication. In Daubert, Dr. Palmer's testimony based on differential diagnosis was also rejected as proof of causation. Judge Kozinski cited the Sixth Circuit's evaluation in Turpin of Palmer's testimony as lacking any scientific basis. He further stated that the differential diagnosis in that case seemed to be based only on a temporal association between the birth defects and Bendectin use rather than a ruling out of other possible causes.⁽¹⁴⁾

     Courts have also attacked differential diagnosis where it rules out other potential causes of disease, but fails to "rule in" the alleged cause of the injury, where there is insufficient scientific evidence to assert general causation.⁽¹⁵⁾ In Schmaltz v. Norfolk and Western Railway,⁽¹⁶⁾ a clinician's testimony was attacked for failing to show that the substance in question could cause the illness claimed even if other substances had been ruled out as causes.

     Other courts have attacked specific differential diagnoses as based on information that cannot justify the conclusions drawn by the physician. In O'Connor v. Commonwealth Edison,⁽¹⁷⁾ the Seventh Circuit held that a physician who testified that visual examination of the plaintiff's eyes indicated that his cataracts were caused by radiation exposure could not have made that determination with the methodology used.⁽¹⁸⁾ Generally, some commentators have attacked the scientific value of a physician's assertion as speculative and too reliant on individual judgment.⁽¹⁹⁾

Could a Differential Diagnosis Provide Evidence Of Causation Where Epidemiology Shows Less Than a Doubling of Background in a Test Population?

A. The Case for Admission
     Epidemiology showing less than a doubling in the background rate of disease can be viewed as a starting point for differential diagnosis. In re Joint East and South District Asbestos Litigation⁽²⁰⁾ provides an example. The claim of a causal connection between asbestos and colon cancer was bolstered by a physician using epidemiological studies showing a ratio between exposed and general population incidence of disease falling between 1 and 2, a correlation possibly indicating causation.⁽²¹⁾ The physician used the tools of differential diagnosis to rule out other causes, especially age and unhealthy diet, which account for much of the background incidence of colon cancer, and backed up his assertion of general causation with reference to research connecting asbestos to other types of cancer. His medical opinion that the disease was caused by asbestos was ruled admissible. Logically, the physician's approach was a qualitative reevaluation of the background rate of disease and the general population sample it was based on in relation to his patient's risk factors. In cases like this, further epidemiological investigation may be too costly, or impossible because of statistical error from a too small sample size.

B. The Case Against Admission
     The physician using epidemiology to back up differential diagnosis cannot be certain of causation, simply some correlation. He also cannot often quantitatively isolate the effects of risks like age or poor diet from one such as asbestos exposure by independently examining the effect of one of those variables on cancer rates.⁽²²⁾ The only certain way to determine whether individuals similar to the plaintiff experience an elevated rate of cancer would be an epidemiological study comparing populations of young, healthy eating people who were and were not exposed to asbestos. It can be argued that allowing any other approach to the information might encourage the sort of "conjecture" rejected by the Ninth Circuit in Daubert.⁽²³⁾

Assessment
     The debate over the admissibility of differential diagnosis operates at two levels. One is the question of whether it's good science, based on time-tested and commonly used methods, or unreliable, qualitative, and based on physicians' best guesses hopefully made in good faith. On another level, this is a question of what is to be done with cases that fall between what is believable, perhaps even probable, and that which is scientifically proven.⁽²⁴⁾ Admission brings the fear of juries unjustly finding liability on the basis of quackery. Exclusion threatens that plaintiffs will be unjustly denied recovery because science is not yet advanced enough to prove their claim, or too expensive for them to access.

1. Dorland's Illustrated Medical Dictionary 461 (27th ed. 1988).
2. Margarate Berger, Federal Judicial Center Manual on Scientific Evidence, Evidentiary Framework, 56-67 (1994); Troyen Brennan, Causal Chains and Statistical Links, 73 Cornell L. Rev. 469 (1988) .
3. Charles Nesson, Agent Orange Meets the Blue Bus: Factfinding at the Frontier of Knowledge, 66 B.U. L.Rev. 521, 526 (1986).
4. Merrel Dow v. Havner, 907 S.W.2d 535, 548 (Tex.App.-Corpus Christi, March 17, 1994), rehearing en banc (Aug. 10, 1995), rehearing overruled (Sept. 28, 1995).
5. This logical approach is characterized as post hoc ergo propter hoc, roughly translated as "after this, therefore because of this."
6. Ellen Relkin, Some Implications of Daubert and its Potential for Misuse: Misapplication to Environmental Tort Cases and Abuse of Rule 706a Court Appointed Experts,15 Cardozo L. Rev. 2255 at 2256.
7. Nesson, supra note 3.
8. Relkin, supra note 6 at 2256.
9. Benedi v. McNeil-P.P.C. Inc., 66 F.3d 1378, 1384 (4^th Cir. 1995).
10. Id. at 1378.
11. Id. at 1378.
12. McCullock v. H.B. Fuller, 61 F.3d 1038 (2d Cir, 1995).
13. Becker v. National Health Products, 896 F.Supp. 100 (N.D.N.Y. 1995).
14. Daubert v. Merrell Dow, 43 F.3d 1311 (9th Cir. 1995).
15. Cavallo v. Star, 892 F.Supp 756, 771.
16. Schmaltz v. Norfolk and Western Railway, 878 F.Supp. 1122 (N.D.Ill. 1995).
17. 13 F.3d 1090 (7th Cir. 1994) cert. den. 114 S.Ct. 2711 (1994).
18. Id. at 1106.
19. Brennan, supra note 2; Ronald Allen, Comment: Rationality, Mythology, and the "Acceptability of Verdicts" Thesis, 66 B.U. L. Rev. 541 (1986).
20. 52 F.3d 1124 (2d Cir., 1995).
21. This ratio is referred to as "Relative Risk."
22. Many substances act synergistically in the body, where the effect together of two agents is greater than the effect of each of them independently when summed. As an example, tobacco smoke and asbestos act synergistically in causing higher rates of some lung cancers than their individual added effects would indicate.
23. 43 F.3d 1311.
24. Nesson, supra note 3 at 526.