A Qualitative Study of Pathologists' Attitudes Toward Patient-Pathologist Interactions

Cathryn J. Lapedis, MD, MPH; Lauren Kroll-Wheeler, MD; Melissa DeJonckheere, PhD; Dustin Johnston; Scott R. Owens, MD


Am J Clin Pathol. 2021;156(6):969-979. 

In This Article

Materials and Methods

Participants and Sampling

This survey was deemed exempt and not regulated by our institutional review board as part of a quality improvement project (HUM00144372). Pathologists were recruited through our university's Department of Pathology Twitter handle (@UMichPath), which linked to a survey produced using Qualtrics survey software. Five tweets were posted over 3 weeks, from March 15 to April 20, 2019, for the approximately 5,800 followers of our university's pathology Twitter feed. The university's pathology Twitter feed is designed and cultivated for pathologists, with daily diagnosis quizzes and other specific, pathologist-centered resources. Initial tweets were timed to correspond with a large international conference for pathologists, when many pathologists were accessing Twitter and following our university's Twitter page. The survey software recorded respondents' IP addresses and location to ensure unique survey responses. All responses were collected in English.

The goal of our study was to understand the attitudes of pathologists toward interacting with patients by showing patients their slides and discussing their pathology report. In the survey, we asked pathologists, "How interested would you be in meeting with patients to discuss their pathology report and show them a slide of their tissue?" Participants were asked to answer this question, assuming that (1) their time was adequately reimbursed and (2) the primary treating clinicians had already told the patients their diagnosis. The first assumption regarding reimbursement was included because currently no reimbursement structure relates to patient-pathologist interactions. Not including an assumption related to reimbursement would likely have obscured the goal of the study because pathologists may prioritize lab work over patient interaction, the latter of which would take time and provide no direct revenue. The second assumption regarding diagnosis disclosure was included because the typical structure of medicine is that pathologists make a diagnosis, and clinicians, who have an established relationship with their patients, disclose the diagnosis. Having a pathologist who has never met the patient communicate the initial diagnosis, especially in the case of malignancy, could be highly controversial for pathologists and, again, obscure the goal of our study.

Interest level was assessed on a 6-point scale, from "definitely interested" to "definitely not interested." We then asked participants to respond via free text to the open-ended question, "Why?" We also asked pathologists their age, gender, rank (ie, resident, junior attending, midcareer attending, senior attending), and type of practice (ie, community, academic, other). See the supplemental materials for the survey instrument (all supplemental material can be found at American Journal of Clinical Pathology online).


The descriptive questions were cleaned and coded using Stata, version 15 statistical analysis software (StataCorp). Descriptive statistics were calculated for each variable. The interest variable was dichotomized into an "interested group" that contained "definitely interested" and "interested" participants. All other responses were grouped into the "not interested" group. We used bivariate logistic regression to determine whether any of the variables, including age, gender, rank, and type of practice, were significantly associated with interest level.

We used a qualitative thematic approach to analyze the free-text data obtained from the open-ended question designed to capture the attitudes of pathologists who may or may not be interested in meeting with patients to discuss their pathology report and to show them a slide of their tissue. Participants had unlimited space to respond to the question in the text box. All responses were independently read by a team of 2 investigators (C.J.L. and L.K.-W.). One investigator was a pathologist who has a background in public health and has taken courses on qualitative research. The other was a pathology resident whom the first investigator had trained in thematic analysis. The 2 investigators began by reading all responses and creating a codebook for relevant and meaningful codes. They independently coded the first 20 responses, and then met as a pair to resolve differences in coding. They repeated this process in progressively larger groupings of responses. When initial codes had been determined and resolved, the researchers grouped individual codes into broader themes and subthemes to accurately reflect the content of the entire data set. Both researchers agreed on the categorization of specific codes into broader themes and subthemes, and any discrepancies were discussed and resolved. We ensured a rigorous qualitative approach by (1) independently defining codes, resolving discrepancies, and reaching consensus to ensure the internal validity, or accuracy, of the coding; and (2) examining and presenting multiple perspectives in the following results, including disconfirming evidence (ie, comments that diverge from or conflict with a particular theme) that adds nuance to the overall thematic findings. The sampling plan achieved thematic saturation.