Malpractice Concerns, Defensive Medicine, and the Histopathology Diagnosis of Melanocytic Skin Lesions

Linda J. Titus, PhD; Lisa M. Reisch, PhD; Anna N. A. Tosteson, ScD; Heidi D. Nelson, MD, MPH; Paul D. Frederick, MPH, MBA; Patricia A. Carney, PhD; Raymond L. Barnhill, MD, MSc; David E. Elder, MBChB, FRCPA; Martin A. Weinstock, MD, PhD; Michael W. Piepkorn, MD, PhD; Joann G. Elmore, MD, MPH


Am J Clin Pathol. 2018;150(4):338-345. 

In This Article

Materials and Methods

Study Design and Procedures

The present analysis uses data from the Melanoma Pathology Study (M-Path), a study of pathologists who interpret MSLs, including benign nevi, dysplastic nevi, and melanoma. The M-Path study methods have been detailed elsewhere.[4] Briefly, we invited participants from 10 states (California, Connecticut, Hawaii, Iowa, Kentucky, Louisiana, New Jersey, New Mexico, Utah, and Washington). Eligibility criteria included completion of residency training and/or fellowship training, interpretation of MSLs within the previous year, expected continuation of interpreting MSLs for at least the following 2 years, and verifiable address of practice location. Of 450 pathologists contacted to participate in the survey, 110 were not eligible, 39 declined without determination of their eligibility status, and 301 confirmed they were eligible to participate, of whom 207 (68.8%) actively enrolled in the study.

Consenting pathologists completed an online survey that elicited general demographic and professional information, including clinical training, perceptions, and experience, including past involvement in medical malpractice litigation. Pathologists were asked to agree/disagree with the following statements regarding how medical malpractice concerns affect their own practice: "I request additional slides cut from the block," "I recommend additional surgical sampling," "I request second opinions," and "I order additional tests such as IHC or molecular tests" (ie, assurance behaviors). Survey response options for the assurance behaviors were formatted using a 6-point Likert scale (slightly disagree, disagree, strongly disagree, strongly agree, agree, slightly agree). We also asked pathologists to report whether the direction of their diagnosis of MSLs was influenced by concerns about medical malpractice. Three response options were available: influence toward a less severe diagnosis, no influence on diagnosis, and influence toward a more severe diagnosis. A full copy of the M-Path survey has been published previously.[14]

Institutional review board approval for all study procedures was obtained from the University of Washington, Fred Hutchinson Cancer Research Center, Oregon Health & Sciences University, Rhode Island Hospital, and Dartmouth College.

Statistical Analysis

Pathologists' responses to individual survey questions on use of assurance behaviors were classified as yes (slightly agree, agree, strongly agree), indicating use of the assurance behavior, vs no (slightly disagree, disagree, strongly disagree), indicating nonuse of the assurance behavior. Due to small cell sizes, which precluded multivariable modeling, we created a composite variable representing use of at least one of the four assurance behaviors; however, applying the same "yes" and "no" categories used for the individual assurance behaviors proved infeasible due to a zero cell. After exploring possible options, we constructed a composite variable in which "yes" corresponded to "strongly agree" and "no" represented all other response categories combined. This approach, which was included in the multivariable model, produced the most stable (narrowest confidence intervals) and the most conservative results (smallest odds ratio).

With regard to the outcome variable (ie, the direction of influence of MSL diagnosis due to concerns about medical malpractice), most pathologists reported either no influence on diagnosis or being influenced toward a more severe diagnosis; only three pathologists indicated influence toward a less severe diagnosis. Consequently, the primary outcome was dichotomized as "influence toward a more severe diagnosis" vs "no influence"; the latter category included the three pathologists who reported being influenced toward a less severe diagnosis.

We used frequency distributions to display the percentage of pathologists reporting that concerns about medical malpractice had no influence on their MSL diagnosis or influenced them toward a more severe diagnosis (the outcome). Simple (unadjusted) logistic regression models explored pathologist characteristics, including past personal involvement in MSL-related malpractice litigation and the use of assurance behaviors, in relation to the outcome. Variables associated with the outcome at P ≤ .10 in the unadjusted logistic models were included in a multivariable logistic regression model that controlled for potential confounding, thereby identifying variables that were independently associated with the outcome. Correlation matrices were used to identify correlated variables. Firth's bias reduction penalized-likelihood approach was used to obtain estimates of regression parameters, confidence intervals, and P values.[15] All statistical analyses were performed using SAS software version 9.4 (SAS Institute, Cary, NC).