Introduction
Epidemiology consists of various processes that are similar to other health professions. These processes are like nursing process, scientific process, diagnostic process and quality improvement process. In each process, the epidemiologist entirely depends on accurate assessments of various disease states. This assessment occurs in almost all aspects of their lives, and all epidemiologists know that diagnostic test results are fallible. This is the reasons as to why they devise other methods to measure the errors in the results. However, few epidemiologists pay attention to evaluation of diagnostic testing.
Reference- test Bias in Diagnostic- test evaluation: Problem in Epidemiology too
The evaluation of diagnostic tests can be seen to consist of many biases, and such evaluation needs to be compared with a reference standard. This reference standard can be said to discriminate either disease or non-disease states remarkably well. However, there exists a reference bias that results from the fact that extremely few reference standards are perfect. The references bias that results quite all-encompassing and demanding type of bias. Such bias causes sensitivity and specificity estimates that may affect disease classifications.
A major problem with reference- test bias occurs when the acquired new test seems better than the reference standard. Such a belief leads to the perception to unearth explanation that may explain reference- test bias. The issue of intuition can be explained using the analysis done by microbiologists in the year 1990 for the diagnosis of chlamydial disease and other infectious diseases. However, this process had an inherent bias such that it consisted of an over estimation of the results. Such discrepancies in epidemiological analysis have happened in several situations, for example, in the US Food & Drug Administration (Miller 15).
The inherent bias causes a breakdown of communication between lab scientists and statisticians. The laboratory scientist is the person who does the first analysis where he makes no assumptions while the statistics methodologist does the reference testing resulting in discrepancies. Patient- infected- status algorithm (PISA) can be seen to replace the discrepant process in evaluating diagnostic tests. It can be seen to be an intuitive method yet effective in evaluating the diagnostic results. However, it must be noted that PISA also comes up with biased results.
For example, two specimens may be used such that each specimen originates from a different anatomic site. However, it must be emphasized that PISA cannot be used as an acceptable process in FDA evaluation since it remains associated with a bias.
Conclusion
The above explanation shows that the intuitive methods that can be used in evaluating diagnostic test consist of biases. This drives the need and desire to get more sophisticated approaches of evaluating diagnostic test results. There are other alternative methods that got developed such as Latent- class analysis and Bayesian approaches which attempt to give advanced results. Such statistical approaches can be seen to be positive steps forward, but the unfortunate thing is that none of them is ideal. The past decades were filled with attempts to progress in evaluation of diagnostic testing, but most mainstream epidemiologists remain out of the discussions. It is vital for mainstream epidemiologists to pay attention to methods used in evaluating diagnostic tests since their studies must be based on diagnostic tests. This is the reason as to why epidemiologists should strive to acquire a deep understating of the diagnostic testing field.
Works cited
Krieger, Nancy. Epidemiology: Why Epidemiologists Cannot afford to ignore Poverty, 2007. Web.
Miller, William C. Commentary: Reference-test Bias in Diagnostic-test Evaluation: A Problem for Epidemiologists, too, 2002. Web.