Use of human epidemiology studies in proving causation.

• Author: See, Andrew

Epidemiology studies can be used to establish only general causation, but there are many criteria that can be employed to test those studies

EPIDEMIOLOGY studies examine populations--generally smaller subsets or samples of populations--in an attempt to determine whether there is an association between exposure to a substance or factor and subsequent disease or injury. These studies are commonly used in the courts to support or refute claims that exposure to a substance or factor caused a disease or an injury in a specific plaintiff or group of plaintiffs.

The problems are (1) on what issues are epidemiological studies properly used and admitted in evidence, and (2) how a particular study was carried out and effect that may have on its admissibility.(1)

USE AND ADMISSION

1. Relevance

   Epidemiology studies are relevant only to general causation. To satisfy the burden of proving causation, plaintiffs must show both (1) general causation--that is, whether the exposure or substance is capable of causing the alleged disease or injury, and (2) specific causation--that is, whether the exposure or substance actually caused the disease or injury in the specific case at issue.(2) Results from epidemiology studies are relevant only to the issue of general causation and cannot establish whether an exposure or factor caused disease or injury in a specific individual.(3) Unless general causation has been established evidence of specific causation in an individual plaintiff is not relevant and not admissible.

2. Necessity

   Human epidemiology studies are a primary and generally accepted methodology to investigate the existence of a causal relationship between exposure to a factor and subsequent disease or injury. They have been held to be the most useful and conclusive type of evidence and that they must be taken into account.(4) Depending on philosophical views and the facts of particular cases, courts have made varied pronouncements regarding the necessity of epidemiology studies in causal analysis.

   Many courts have held that it is necessary to offer epidemiology evidence to prove causation.(5)

   Courts have held that epidemiological studies are necessary to prove causation, but only if there is no biological evidence of a causal mechanism for the disease. For instance, in Daubert v. Merrell Dow Pharmaceuticals Inc. at the district court level, the court stated: "Absent a scientific understanding of the cause of [the injury or disease], causation may be shown only through reliance upon epidemiological evidence."(6)

   Courts also have held that epidemiology studies are valuable but not dispositive of the causation issue. Epidemiology studies do not always prevail over non-epidemiological evidence.(7)

   In inexplicable rulings that seem to beg the question of the necessity of epidemiology studies, a few courts have stated that epidemiology evidence should be required only when they are available or if it is possible to gather the data.(8) The far better view, and that consistent with the Supreme Court's directions in Daubert, is that tort cases must be decided on the basis of the best available scientific evidence at the time of trial. Dispute resolution in the tort system cannot wait or be undermined by the creation of lower standards simply because epidemiology evidence is not currently available. As the Fifth Circuit said in Moore v. Ashland Chemical Co.: "In sum, the law cannot wait for future scientific investigation and research. We must resolve cases in our courts on the basis of scientific knowledge that is currently available."(9)

   To round out all permutations, some courts have flatly held that epidemiology evidence is not required to prove causation in an exposure case.(10)

   CONDUCT

   Epidemiology studies must be conducted properly and utilized in the legal context following the scientific method. This approach contemplates at least five steps in the reasoning process:

   * Formulation of the hypothesis or research question;

   * Preparation of a written protocol or study plan that provides, in advance of gathering any data, the methodology to be followed in the study;

   * Execution of the protocol, that is, gathering data from the population specified using the methodology prescribed in the research plan;

   * Analysis of the data using the statistical methodology prescribed in advance in the protocol; and

   * Drawing permissible conclusions, if any, from the study results, including determining whether the study results ought to be utilized at all in forming conclusions or generalizing findings to the population at large.

   These steps apply both to the critique of any specific epidemiology study and the assessment of its evidentiary value, if any, as well as to the evaluation of an expert witness's proper or improper use of study data to support a causation opinion.

1. Hypothesis

   The hypothesis or research question must be set in advance of conducting the study, before the investigator knows what the data will show. These questions usually are framed in the negative--for example, that a particular form of cancer is not associated with exposure to chemical A. The investigator then attempts by the study to disprove the hypothesis by showing that the disease is, in fact, associated with A. Both as study investigators and expert witnesses, investigators are required by the scientific method to be objective in gathering data or in preparing a causation opinion--not knowing how the opinion will come out based on what the data will show.

   Forming a conclusion before research is conducted is contrary to the scientific method.(11)

2. Protocol

   In advance of any data gathering, there must be a written protocol or research plan setting out the study's methodological steps. The protocol should include all information regarding the hypothesis, the selection of the study population, how and with what measuring instruments the data is to be gathered and how it is to be analyzed. All assumptions the investigator has made should be fully disclosed in the protocol and reviewed critically. If the investigator has made assumptions that are epidemiologically unsound or contrary to common sense, there is a great potential that the study results may be adversely affected.

   The end points (that which is observed and measured in the course of the study) chosen by the investigator should be critically evaluated to ensure that the study is looking at symptoms that are relevant and make scientific and medical sense. For example, in Cowen v. Mobil Oil Corp.,(12) a study looked at only mortality rates among employees and concluded that a work place exposure resulted in no excess deaths. The court criticized the study because, among other things, it did not reevaluate the morbidity rate, that is, the frequency of illnesses not resulting in death, of the study group.

   There are two basic kinds of epidemiological studies: cohort studies and case control studies. A cohort study examines the incidence or frequency of disease in an exposed group and compares it to the frequency of disease in an unexposed control group. A case control study matches subjects who already have the disease with a control group of subjects who do not have the disease. The frequency of the exposure to the substance or factor is then compared between the groups.

   In order to determine whether there is an association between the exposure or factor and subsequent disease or injury, the exposed group must be compared to an unexposed control group or population.(13)

3. Execution

   1. Biases

      Both the methodology and the actual execution of data gathering must be evaluated to ensure that study subjects are similar regarding their baseline characteristics and that one or more biases have not been injected into the data. There are many types of biases that routinely affect study results.

      (a) Selection Bias

      Selection bias occurs when the exposed group is selected in a way that makes it more or less susceptible to the disease for reasons that are independent of the exposure. In the breast implant litigation, the court stated that a valid epidemiological study requires that subjects be chosen by an unbiased sampling method from a definable population.(14)

      (b) Recall Bias

      Subjects exposed to the substance or factor are more likely to recall and relate symptoms to their exposures than are those not exposed. It is important to "blind" the questioner or survey taker--that is, make the questioner unaware of which subject is exposed or unexposed to remove the danger that the questioner will probe harder for information about symptoms among the exposed subjects. "Blinding" to the exposure status of the study subjects is an important methodology to eliminate recall or information bias.(15)

      (c) Ascertainment, Detection or Diagnosis Bias

      Both the methodology and the execution of data gathering must be evaluated to determine whether disease is being detected or diagnosed at similar rates between comparison groups. Blinding the investigators to the exposure status of the study subjects is an important technique to avoid such bias. It also is important to ensure that uniform and consistent rules and criteria are applied to what questions are asked and how a disease is diagnosed or how a symptom is characterized to ensure that information about the presence of a disease is consistently interpreted across different investigators and different study subjects. The skill and training of the observers and appropriate quality controls to insure...