TY - JOUR
T1 - An analysis reveals differences between pragmatic and explanatory diagnostic accuracy studies
AU - Bossuyt, Patrick M.
AU - Olsen, Maria
AU - Hyde, Chris
AU - Cohen, J. rémie F.
PY - 2020/1
Y1 - 2020/1
N2 - Objectives: The objective of this study was to clarify a difference between two approaches while evaluating the diagnostic accuracy of medical tests, labeled here as “pragmatic” vs. “explanatory” studies. Methods: Using the definitions and characteristics described by Schwartz and Lellouch for randomized trials of interventions, and Schwartz' more general distinction between a pragmatic and an explanatory approach in medical research, we define a similar continuum for diagnostic accuracy studies. Explanatory studies aim to better understand the behavior of a test; pragmatic ones are done to support recommendations or decisions about using the test in clinical practice. Results: Pragmatic test accuracy studies differ from explanatory test accuracy studies in several ways. The difference in aims has implications for key elements of study design, such as the study eligibility criteria, the recruitment of patients, the reference standard, and the choice of the statistical analysis. Explanatory accuracy studies are often designed to test a hypothesis. They are typically selective in recruitment, may include “healthy controls,” with a small sample size, often recruited at a single center. They ignore testing failures in the analysis and more often present their results as ROC curves. By contrast, pragmatic studies are designed to guide decision making. They ideally will recruit a single, large, and representative group of patients at multiple sites and will more often present their results as estimates of sensitivity and specificity or predictive values at a prespecified threshold. Conclusion: Distinguishing between a pragmatic and an explanatory approach can help in the design, analysis, and interpretation of diagnostic accuracy studies. It can clarify debates about the appropriateness of design features to the study purpose and about the validity and applicability of study findings.
AB - Objectives: The objective of this study was to clarify a difference between two approaches while evaluating the diagnostic accuracy of medical tests, labeled here as “pragmatic” vs. “explanatory” studies. Methods: Using the definitions and characteristics described by Schwartz and Lellouch for randomized trials of interventions, and Schwartz' more general distinction between a pragmatic and an explanatory approach in medical research, we define a similar continuum for diagnostic accuracy studies. Explanatory studies aim to better understand the behavior of a test; pragmatic ones are done to support recommendations or decisions about using the test in clinical practice. Results: Pragmatic test accuracy studies differ from explanatory test accuracy studies in several ways. The difference in aims has implications for key elements of study design, such as the study eligibility criteria, the recruitment of patients, the reference standard, and the choice of the statistical analysis. Explanatory accuracy studies are often designed to test a hypothesis. They are typically selective in recruitment, may include “healthy controls,” with a small sample size, often recruited at a single center. They ignore testing failures in the analysis and more often present their results as ROC curves. By contrast, pragmatic studies are designed to guide decision making. They ideally will recruit a single, large, and representative group of patients at multiple sites and will more often present their results as estimates of sensitivity and specificity or predictive values at a prespecified threshold. Conclusion: Distinguishing between a pragmatic and an explanatory approach can help in the design, analysis, and interpretation of diagnostic accuracy studies. It can clarify debates about the appropriateness of design features to the study purpose and about the validity and applicability of study findings.
KW - Diagnostic accuracy
KW - Explanatory studies
KW - Pragmatic studies
KW - Reference standard
KW - Sensitivity
KW - Specificity
KW - Test evaluation
UR - http://www.scopus.com/inward/record.url?scp=85073508323&partnerID=8YFLogxK
U2 - https://doi.org/10.1016/j.jclinepi.2019.09.017
DO - https://doi.org/10.1016/j.jclinepi.2019.09.017
M3 - Article
C2 - 31561014
SN - 0895-4356
VL - 117
SP - 29
EP - 35
JO - Journal of Clinical Epidemiology
JF - Journal of Clinical Epidemiology
ER -