top of page
  • Writer's pictureTeamResearch

Validating/Reliability Testing Data/Three Different Disease Diagnoses of the DOGRISK Questionnaire

Updated: May 9, 2020

ARTICLE DATE: February 2016


CITATION: Roine J, Uusitalo L, Hielm-Björkman A. Validating and reliability testing the descriptive data and three different disease diagnoses of the internet-based DOGRISK questionnaire. BMC Vet Res. 2016;12:30. Published 2016 Feb 20.



The DOGRISK questionnaire is an internet-based ongoing study of canine nutrition, living environment, and disease. Here we aim to assess the performance of the questionnaire using data from the first three years in relation to some descriptive and disease variables. We used associated questions, official register records, test-retest repeatability, and email/mail contact with questionnaire respondents.


Reliability against an official register of gender, season of birth, breed, and results of hip radiography was tested and Cohen's Kappa was between 0.95-0.99. Internal consistencies of hypothyroidism status and dog's age were calculated using Cronbach's Alpha (α = 0.95 and α = 0.99, respectively). Test-retest repeatability of ten variables among 224 participants was analyzed. Gender, season of birth, and born in owner family had Cohen's Kappa > 0.86, color of coat, vaccination status as an adult, other dogs in household, and vaccination status as a puppy had Cohen's Kappa between 0.67-0.80, and body condition score under two months of age and tidiness of household, had Cohen's kappa of 0.45 and 0.42, respectively. In addition, time spent outside had Cohen's kappa of 0.37. Of the owners contacted by email/mail to confirm their dog's atopy/allergy (skin symptoms), 8.9% reported that they had given an incorrect answer (positive predicted value 91%), but only 69% of all reaffirmed positive answers had a diagnosis set by a veterinarian.


Our study showed that owners were diligent with basic information and with the status of three diseases. Cohen's Kappa in the reliability of the test-retest was in most variables at least 0.67. We propose that the descriptive variables and the disease variables be used as such when we generate hypotheses from the DOGRISK data.


bottom of page