Table 1.
Conformational measurements taken through soft tape and photographic measurements.
Fig 1.
Photographic conformational measurements.
(A) Eye width ratio calculated from frontal plane photographic measurements (intercanthal distance (pink line) and skull width (green line)). (B) Craniofacial ratio calculated from sagittal plane photographic measurements (muzzle length (red line) and cranial length (blue line)). Reprinted under a CC BY license, with permission from Ali Limentani, original copyright 2024.
Fig 2.
P values of pairwise comparisons between the proportion of Grade 0 dogs in brachycephalic breeds that had a p value >0.005 when compared to Pugs, French Bulldogs or Bulldogs.
Adjusted p-values of one-sided Fisher’s exact test to test the alternative hypothesis that the fourteen brachycephalic breeds in this study have a greater proportion of dogs scoring Grade 0 in the respiratory function assessment. All other breeds studied (Affenpinscher, Boxer, Cavalier King Charles Spaniel, Chihuahua, Maltese, Pomeranian, Staffordshire Bull Terrier) had significantly greater proportion of Grade 0 dogs in the sample population with p-values of 0.005 against all 3 breeds.
Fig 3.
BOAS grade distribution across the breeds tested in comparison to the 3 most popular breeds.
Pug, French Bulldog and Bulldog grade distributions taken from previous study (18). (A) High risk breeds: Pekingese (n = 45) and Japanese Chin (n = 46). (B) Moderate risk breeds: Griffon Bruxellois (n = 52), Boston Terrier (n = 107), King Charles Spaniel (n = 83), Dogue de Bordeaux (n = 51) and Shih Tzu (n = 42). (C) Mild risk breeds: Staffordshire Bull Terrier (n = 120), Cavalier King Charles Spaniel (n = 73), Chihuahua (n = 47), Boxer (n = 79), Affenpinscher (n = 69), and Pomeranian (n = 51). Maltese data not included in this figure as number of affected dogs insufficient to draw reasonable conclusion (total n = 32, Grade 1 n = 1).
Table 2.
Multiple logistic regression analysis final model. Body condition score [1]: ≥ 6 compared to reference level [0]: ≤ 5. Nostril stenosis [1]: mild stenosis, [2]: moderate stenosis and [3]: severe stenosis compared to reference level [0]: open nostrils.
Fig 4.
Final model from the multiple logistic regression analysis incorporating BCS, nostril stenosis and CFR.
(A) Predicted probability of the model correctly classifying Grade 0 (Observed 0) or Grade 1−3 dogs (Observed 1). Negative predictive power of 71% and positive predictive power of 69%. (B) Area under ROC curve = 0.75 (95% CI: 0.72-0.79).
Fig 5.
Percentage distribution of body condition score by BOAS Grade five breeds with statistical significance.
P-values indicate the raw p-value from the two-sided Fishers-exact test comparing normal or underweight (BCS ≤ 5) to overweight (BCS ≥ 6). Included are breeds with a statistical significance from the multivariate breed-specific models (Affenpinscher and Shih Tzu).
Fig 6.
Percentage distribution of nostril stenosis severity across different brachycephalic breeds by BOAS Grade.
P-values indicate the raw p-value obtained in univariate analysis by two-sided Fisher’s exact test for binary outcomes (CKCS, Pomeranian and SBT) or Cochrane-Armitage test for trend (remaining groups).
Fig 7.
Craniofacial ratio values across the fourteen brachycephalic breeds included in the study.
(A) Mean craniofacial ratio in each breed comparing Grade 0 to Grade 1-3 dogs. Error bars indicating standard deviation. Significant results annotated *. (B) Mean craniofacial ratio for each breed plotted against the percentage of Grade 0 dogs within that breed. Error bars indicate standard deviation.
Fig 8.
Simple logistic regression model plotted from the fourteen brachycephalic breeds indicating the probability of BOAS grades at different craniofacial ratios.
(A) Craniofacial ratio plotted against the probability of being BOAS Grades 1−3 versus Grade 0. Grey area indicates the 95% confidence interval. (B) Receiver Operator Characteristic (ROC) curve. Area under ROC curve = 0.74 (95% CI: 0.70-0.77, p < 0.0001).