Common Pitfalls in Analysis of Tissue Scores

Abstract

Histopathology remains an important source of descriptive biological data in biomedical research. Recent petitions for enhanced reproducibility in scientific studies have elevated the role of tissue scoring (semiquantitative and quantitative) in research studies. Effective tissue scoring requires appropriate statistical analysis to help validate the group comparisons and give the pathologist confidence in interpreting the data. Each statistical test is typically founded on underlying assumptions regarding the data. If the underlying assumptions of a statistical test do not match the data, then these tests can lead to increased risk of erroneous interpretations of the data. The choice of appropriate statistical test is influenced by the study's experimental design and resultant data (eg, paired vs unpaired, normality, number of groups, etc). Here, we identify 3 common pitfalls in the analysis of tissue scores: shopping for significance, overuse of paired t-tests, and misguided analysis of multiple groups. Finally, we encourage pathologists to use the full breadth of resources available to them, such as using statistical software, reading key publications about statistical approaches, and identifying a statistician to serve as a collaborator on the multidisciplinary research team. These collective resources can be helpful in choosing the appropriate statistical test for tissue-scoring data to provide the most valid interpretation for the pathologist.

Fundamental Concepts for Semiquantitative Tissue Scoring in Translational Research

“…It is not appropriate to use parametric analysis to analyze data derived from ordinal scoring systems. 3,4,46 Evaluation…”

“…Once the semiquantitative tissue scores are collected, appropriate statistical tests can be applied; these have been reviewed. 4,5,8,45,46 As mentioned above, appropriate expertise such as a statistician collaborator would be advantageous to guide proper statistical analyses of the data. Awareness of the type of data produced by semiquantitative scoring is very important because it guides the type of statistical tests used to give the most compelling interpretations of the study.…”

Histopathologic Evaluation and Scoring of Viral Lung Infection

“…This approach, which lacks the expertise of a pathologist trained in tissue interpretation, has been labeled as "do-it-yourself pathology" and is linked to multiple publications containing erroneous interpretations [22,25,31,32]. While observations made by biomedical personnel may be biologically accurate in some cases, it is important to note that tissue examination by non-pathologists (even those who are "scientific experts" for a particular disease) is prone to false-positive and false-negative errors and not recommended [33]. Ideally, tissues should be examined by a pathologist familiar with histopathology of the model (see Note 1).…”

“…If the ordinal scoring for seven mice per group produced the following results for WT (3, 3, 2, 3, 4, 3, 4) and mutated mice (1, 2, 1, 1, 1, 1, 2), then the data can be statistically analyzed. Importantly, ordinal scores do not meet the assumptions required for parametric tests; thus nonparametric tests should be used [33]. For the mock example, the difference between groups using a Mann-Whitney U-test was significant (P ¼ 0.002).…”

“…Inappropriate use of paired t-tests and shopping for significance are two issues that have slipped into the published literature and potentially compromise the interpretation and reproducibility of studies [33]. For the various scoring methods, statistical analyses of the data should involve the collaborative expertise of a statistician to be able to identify the most relevant tests to confidently evaluate for group differences [22,33,52,53].…”