Abstract
Keywords
The previous and current issues of Veterinary Pathology, September and November 2018, include 2 commentaries on designing and conducting observational studies in veterinary pathology. 1,2 In this editorial, I offer 10 suggestions and critiques of current practices. These reflections arise from assessing or editing about 1400 manuscripts submitted to Veterinary Pathology over the past 5 years. Each of these recommendations is discussed in detail in the accompanying commentaries. 1,2
Work on something important. Consider the major problems facing animal and human health and how we can contribute to solving them. What are the fundamental gaps in knowledge of animal diseases, and how do our skill sets and resources allow us to address these gaps? At times, veterinary pathologists’ fascination for the microscopic carry over to our outlook on science, tempting us to investigate trivial minutiae while skirting the big issues that impact today’s world. We have much to offer, and we should tackle important problems.
Focus observational studies on a specific and testable hypothesis, major question that can be definitively answered, or specific objective that will be completely met when the investigation is done. This is a serious criticism of studies submitted to Veterinary Pathology, where the study objectives are often vague and seem to be written after the study is underway (or worse, as the manuscript is being written). A study based only on a novel method is unlikely to have impact unless it is framed by a compelling rationale and a clear hypothesis, question, or objective.
Elementary school children know that hypothesis testing is fundamental to the scientific method, yet hypotheses are rarely sighted in the pages of Veterinary Pathology. Some of our studies are exploratory or descriptive, and it is a mistake to force those round pegs into the square hole of hypothesis testing. However, for many studies published in the journal, our science will be stronger if we return to the basics of the scientific method to frame observational studies around a clear and precise hypothesis built on a solid rationale.
Creativity inspires innovation and discovery. Discovery is not simply a matter of cataloging one’s observations but requires the creative development of novel ideas, an innovative investigation, thoughtful reflection and insight, and critical analysis. Take studies beyond a mundane description of lesions and their frequency. Investigators should use their imagination, ideas, innovation, and insight to develop a study that reveals something sparkling and new.
Consider various study designs. For some descriptive studies, adding a control group to create an analytic study will allow inferences that are not otherwise possible. If the objective of your study is to “describe”, try changing it to “compare” and consider how this might energize the study. Compare the strengths and limitations of cross-sectional, case-control, and cohort designs and contemplate which is best suited to the needs of the project. There are many observational study designs and approaches to selecting study groups; the most obvious strategy is not always best suited to conclusively address the hypothesis or objective of the study.
Clearly define the study population. This includes the nature of the population being investigated, inclusion and exclusion criteria, and differences between cases and controls with respect to factors that could bias the results. A basic aspect of describing the methods is the details of how cases and controls were selected. For too many studies in veterinary pathology, the cases seem to have been selected from a pensieve holding the memories of the investigators and then compared to a haphazardly selected group of more recently encountered normal animals that bear little resemblance to the case population. Cases and controls should be comparable in all respects except the factor being studied. Clarify this by thoughtfully defining the study population using precise inclusion and exclusion criteria and deliberately considering the potential for bias and confounding.
Raise n to a higher level. Do a sample size calculation, even if it is a crude estimate. Add more cases by expanding the timeframe of sampling, modifying the inclusion and exclusion criteria, and collaborating with other institutions. For almost all observational studies published in Veterinary Pathology, the results are constrained by low numbers of study subjects. Is it better to spend our time and money on more testing for each animal (better for exploration) or adding more animals to the study (better for hypothesis testing and conclusively answering questions)?
Consider the sequence of causation. If the study measures an association between 2 parameters in a single sample, will it be possible to determine the causal sequence? Will the results be valuable if we cannot distinguish which is the preceding cause and which is the later result? In studies based on archived samples, authors are often required to apologetically acknowledge this limitation. Give thought to designing a study that can more conclusively address causality.
Understand the concepts of bias and confounding as these can easily affect the results of observational studies of archival laboratory samples. Carefully consider how bias might occur in a study and what can be done to mitigate it. These exercises in logic and causal reasoning are mentally taxing for mathematically disinclined pathologists but essential for valid results.
Banish the sin of imprecise, subjective, and vaguely defined diagnostic criteria, particularly for grading schemes intended to be used by others. Such criteria introduce interobserver variability, make it impossible for pathologists to reliably apply the study findings to their daily practice, and cloud the comparison and replication of research studies. Compare tumor grades among expert and nonexpert observers. Variability in the scores would indicate a need for more objective and precisely defined criteria.
Validate new methods. The lab is more subtle than the kitchen: a flawed cookie recipe will be immediately recognizable, but spurious laboratory results may be difficult to detect if investigators are not committed to validating assays before their use in a study. The issue of antibody specificity for use in non–target species seems to be particularly ignored in immunohistochemistry studies by veterinary pathologists. Report the results of assay validation (at least as supplemental materials) if you expect reviewers and readers to believe the study findings.
Observational studies in veterinary pathology are the principal basis for current practices in diagnostic pathology and for much of our knowledge of the morphology and pathogenesis of animal diseases. To ensure that observational studies continue to advance knowledge in our discipline, veterinary pathologists should investigate important problems affecting animal or human health, develop an imaginative and innovative approach, use valid study design and methods, find something new, critically analyze the results, and communicate the findings effectively.
