2005 Scientific Session Abstracts

Terminology Used to Express Diagnostic Probability in Cytopathology: Interpretation by Veterinary Clinicians and Effect on Clinical Decision-Making

Mary M. Christopher, DVM, PhD (mmchristopher@ucdavis.edu) 1; Christine S. Hotz, DVM, MS 2; Sonjia M. Shelly, DVM 3; Paul Pion, DVM 4. 1School of Veterinary Medicine and 2School of Medicine, University of California-Davis; 3IDEXX Veterinary Laboratories, West Sacramento, CA; 4Veterinary Information Network, Davis, CA

Context : Fine-needle aspiration cytology often is used to obtain a diagnosis of lymphoma in dogs. When chemotherapy is not an option, the client may elect euthanasia for their pet. In a previous study, veterinary clinical pathologists were found to use a wide variety of terms to express the probability of a cytologic diagnosis. Such terms have imprecise meanings that may affect interpretation and clinical decisions.

Technology : A web-based survey was provided during Sept-Dec 2004 to veterinary clinicians enrolled in the Veterinary Information Network. Results were analyzed statistically using ANOVA following variance stabilization and nonparametric techniques.

Design : The survey included: A) demographic information; B) assignment of percentage probabilities to 18 modifiers of a positive diagnosis of lymphoma; and C) selection of a “most likely” clinical action based on diagnoses modified by probability terms. Results for (B) were compared with those obtained previously from veterinary pathologists.

Results : Respondents were veterinarians in general practice (86.4%), specialty practice (9.4%), and veterinary schools (4.2%).

Table. Probabilities assigned to modifiers by veterinary clinicians compared with pathologists.

Cytologic Diagnosis

Clinicians (n = 740)

Pathologists (n = 90)

P

Median (10-90 thpercentile)

Median (10-90 thpercentile)

Diagnostic for lymphoma

99 (80-100)

100 (95-100)

.0001

Consistent with lymphoma

80 (60-99)

85 (70-95)

.0015

Probable lymphoma

75 (50-90)

80 (60-90)

.1463

Suggestive of lymphoma

75 (50-90)

75 (60-88)

.1230

Possible lymphoma

50 (40-75)

50 (34-75)

.3358

No evidence for lymphoma

10 (0-90)

5 (0-38)

.0071

Clinicians were 3X more likely to initiate treatment with a diagnosis of consistent with- as compared to probable-; and 31X more likely to initiate treatment with a diagnosis of probable- as compared to possible-. For clients electing euthanasia if the diagnosis was cancer, clinicians were 2X as likely to recommend euthanasia with a diagnosis of consistent with- as compared to probable-; and ~18X as likely to recommend euthanasia with a diagnosis of probable- as compared to possible-. About 12% of clinicians indicated they usually ignore any descriptive terms altogether.

Conclusions : Veterinary clinicians are highly variable in their interpretation of probability expressions. Probability terms have a major impact on clinical decision-making, including the decision to euthanize pets.