![]() Sometimes experimental variation nonetheless remains. Replication of the experiment reduces the likelihood that some chance factor was systematically different between the two experimental arms. For example, a researcher would make all conditions (dilution protocols, incubators, etc.) identical between the two conditions except for the variable of interest (i.e. The first is to attempt to eliminate unwanted differences between the compared groups (in the design) and to measure and account for any unavoidable differences (in the analysis). For example, perhaps there is a contaminant in the cytokine preparation that directly kills bacteria, or perhaps the cytokine itself kills bacteria, or perhaps some other unintended difference between the treated and untreated conditions (e.g., temperature or pH) caused the differential survival of the bacteria.Įach of these unintended differences is broadly similar to a confounder – a characteristic associated with the exposure (presence or absence of the cytokine) and causes the outcome (differences in bacterial counts), thereby causing a spurious association between the presence of the cytokine and differences in bacterial counts.Įxperimental biologists address such concerns in two ways. Nonetheless, concern remains that something other than cytokine-aided, neutrophil-mediated killing may be responsible. If the investigator finds fewer live bacteria in condition 1 than in condition 2, the finding is consistent with the hypothesis that the cytokine enhanced neutrophil-mediated killing. After incubation, the bacteria are enumerated and the number of live bacteria compared between conditions 1 and 2. In condition 1, the cytokine is added, and in condition 2, some inert substance such as saline solution is added. 2 An experiment is devised in which neutrophils, bacteria, and growth medium are mixed together. For example, consider the hypothesis that a particular cytokine-a chemical involved in signaling in the immune system-enhances the killing of a species of bacteria by neutrophils, a class of white blood cells. ![]() One might imagine that the experimental method would circumvent most threats to the validity of causal inference that occur in observational studies. Although the particular threats to causal inference are different in experimental and observational sciences, the use of negative controls is a valuable means of identifying noncausal associations and can complement other epidemiologic methods for improving causal inference.Įxperimental biology: threats to causal inference and the use of negative controls We describe the use of negative controls in experiments, highlight some examples of their use in epidemiologic studies, and define the conditions under which negative controls can detect confounding in epidemiologic studies. Biologists employ “negative controls” as a means of ruling out possible noncausal interpretations of their results. Nonetheless, experimental biologists routinely question whether they have correctly inferred causal relationships from the results of their experiments. In experimental biology, the manipulation of experimental conditions prevents many of the noncausal associations that arise in observational studies. The dashed line between L and U indicates that either may cause the other, and they may share common causes. We conclude that negative controls should be more commonly employed in observational studies, and that additional work is needed to specify the conditions under which negative controls will be sensitive detectors of other sources of error in observational studies.Ĭausal diagram for the effect of an exposure of interest (A) on an outcome of interest (Y), with confounders L (assumed measured) and U (assumed uncontrolled) that cause both A and Y. We distinguish two types of negative controls (exposure controls and outcome controls), describe examples of each type from the epidemiologic literature, and identify the conditions for the use of such negative controls to detect confounding. In epidemiology, analogous negative controls help to identify and resolve confounding as well as other sources of error, including recall bias or analytic flaws. We argue, however, that a routine precaution taken in the design of biological laboratory experiments-the use of “negative controls”-is designed to detect both suspected and unsuspected sources of spurious causal inference. Such problems are not expected to compromise experimental studies, where careful standardization of conditions (for laboratory work) and randomization (for population studies) should, if applied properly, eliminate most such non-causal associations. Many techniques have been developed for study design and analysis to identify and eliminate such errors. Non-causal associations between exposures and outcomes are a threat to validity of causal inference in observational studies.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |