Analyzing the analysis

Abstract

Research is often seen as tremendously time- and resource-intensive, difficult to undertake save for the initiated few. In fact, research is actually an uncomplicated undertaking if clearly planned and executed in a straightforward method. The ultimate reward is that of discovery. Proving the expected theory can be gratifying, but nothing comes close to finding something unexpected and then searching for an explanation. It is the unraveling of a mystery which probably ranks as one of the preeminent satisfactions for an academically minded person. Prior articles have laid the foundation of critical appraisal by discussing such topics as study types and bias, random assignment, and the importance of equivalent patient care and adequate follow-up. All these factors profoundly influence the quality and credibility of a research article. It is the analysis which forms the culmination of the research effort. If done right, it can be a major accomplishment; if flawed, it is a wasted effort or even worse—may lead to wrong conclusions with unforeseeable consequences. Analysis is the consideration of everything from description and characterization of the study population to the analytical statistics performed for treatment comparisons. The proper statistical analysis is dependent on the study objectives and design, and the method and source of data collection. Without needing a deep understanding of biostatistics, there are a few basic tenets one can easily evaluate to determine if the appropriate analyses were used when critically appraising an article. Most analyses are divided into descriptive and analytical statistics and should be clearly described in the methods section. Descriptive statistics are most frequently used to provide general information about patients and factors that may be related to outcomes. They set the stage for some of the analytical methods (such as control for confounding) that may be needed to ensure the most accurate estimate of a study effect. Analytical statistics allow for evaluation of treatment effects and the associations between factors. Both may involve tests of statistical significance. The following provides an overview of what to look for when critically appraising a study's analytic methods.