Peer reviews are not certain criteria of a work’s quality. Understanding of study design and systematic error/bias is critical for a researcher. This essay investigates the abstract of a sample study and attempts to determine whether it displays any faults related to these aspects.
Case Study Design
The authors describe their research as a retrospective, cross-sectional, randomized case-control study. However, according to Gordis (2014), “it is unfortunate that the term retrospective has been used for case-control studies” (p. 191). He goes on to state that case-control studies are always retrospective and, therefore, it is not necessary to declare the property explicitly.
Furthermore, the methodology of the study does not correspond with the term “cross-sectional”, as defined by Sedgwick (2014). Sedgwick (2014) describes cross-sectional studies as determining the exposure and outcome at the same time. However, in the sample study the subjects with exposure to phones, as well as those without, were selected before the examinations.
Exposures and Outcomes
The subjects, other than the control group, had been exposed to long-term mobile phone use. The aim of the study was to explore the relation of such use to various hearing issues. While the study did not find a significant difference in any of the parameters, there was some correlation between mobile phone use and high-frequency loss, as well as other abnormalities.
Random Bias and Systematic Error
According to Trochim, Donelly, and Arora (2015), random bias is the error that results from the observation of data. Systematic error, on the other hand, is caused by mistakes made by the researcher. Random bias is mostly unavoidable, while systematic error can be eliminated once the cause is identified and addressed.
Primary Types of Systematic Error
There are three primary types of systematic error: selection bias, information bias, and confounding. Different sources cause each of them, and all three kinds are significant factors to consider during an epidemiological study. Following is a description of each type with some examples.
Selection bias is an error in which exposed and non-exposed individuals are chosen in a way that makes an association appear likely, where in reality there is none. For example, Ammori, Panchani, Gregory, Wylie, and Paul (2015) had to report the results of their twenty-year study with a cautionary note. While the life expectancies of the patients given the surgery were prolonged, only patients whose condition permitted surgical intervention in the first place were selected.
Information bias occurs when the means of collecting data are insufficient, leading to the gathered information being incorrect. Abuabara and Margolis (2015) point out that such a bias may be present in electronic health databases, particularly in those with patient self-reporting. They provide an example of families with a history of melanoma receiving frequent skin checks, leading to the assumption they have other skin diseases.
Lastly, confounding may happen when a factor C, which causes B and is associated with A, surfaces during an investigation of whether A causes B. Glazer et al. (2017) point to such a bias in the studies relating to male infertility and certain cancers. The researcher may assume A causes B when in reality it does not, and most subjects with B are affected by C as well as A.
The study in question may be subject to selection bias, as it took place in a hospital, where people naturally have health issues. It is also subject to significant random bias, as the sample size is rather small. Confounding might be an issue as well since hearing loss can be caused by noise pollution from phone dynamics or earphones.
Conclusion
The study in question has some issues, from improper classification to vulnerability to multiple biases. If a definitive conclusion is to be reached, more extensive experiment with broader coverage and investigation into additional factors will be required. At present, however, the findings of the study do not warrant serious attention.
References
Abuabara, K., & Margolis, D. J. (2015). Databases for clinical research. The Journal of Investigative Dermatology, 135(8), 1-4.
Ammori, M., Panchani, S., Gregory, J., Wylie, J., & Paul, A. (2015). Survival rates following skeletal metastases—A twenty-year analysis. Open Journal of Orthopedics, 5(9), 288-296.
Glazer, C. H., Bonde, J. P., Eisenberg, M. L., Giwercman, A., Hærvig, K. K., Rimborg, S.,… Bräuner, E. V. (2017). Male infertility and risk of nonmalignant chronic diseases: A systematic review of the epidemiological evidence. Seminars in Reproductive Medicine, 35(3), 282-290.
Gordis, L. (2014). Epidemiology. Philadelphia, PA: Elsevier/Saunders.
Sedgwick, P. (2014). Cross sectional studies: Advantages and disadvantages. BMJ: British Medical Journal, 348. Web.
Trochim, W., Donnelly, J. P., & Arora, K. (2015). Research methods: The essential knowledge base. Boston, MA: Cengage Learning.