An assessment of the agreement between the New Zealand Veterinary Association Hip Dysplasia Scoring System and the PennHIP distraction index in German Shepherd dogs

Abstract

AIM: To determine the level of agreement between the New Zealand Veterinary Association (NZVA) Hip Dysplasia Scoring System and the University of Pennsylvania Hip Improvement Program (PennHIP) Distraction Index in German Shepherd dogs, and whether using the NZVA subtotal score or its components affected the level of agreement. METHODS: A prospective study was performed using 47 German Shepherd police dogs undergoing breeding evaluation. All dogs were scored using the NZVA system and the PennHIP index. The relationships between the individual hip-distraction index scores and the scores from the NZVA system, i.e. the total score, the subtotal score, and the scores for the categories making up the subtotal score, were analysed using correlation, followed by univariate ANOVA for the subtotal categories alone. The scores from the NZVA system and the distraction index were then dichotomised into either low or high risk of canine hip dysplasia (CHD). A sign test was then used to determine whether three NZVA thresholds identified the same proportion of ‘at-risk’ dogs as the distraction-index threshold. Where this was the case, the Kappa value was calculated to identify the degree of agreement between the NZVA measures and the distraction index. RESULTS: The left-hip distraction index was significantly correlated to both left-subtotal and left-total NZVA score, however for right-hip scores there was no such correlation. The individual categories of the subtotal NZVA score were not significantly associated with the distraction index except for the subluxation score of the left hip. The proportion of dogs identified as being ‘at risk’ for CHD identified using a distraction-index threshold of 0.3 was similar to that identified by an NZVA total or subtotal score of >2 (44/47 cf. 45/47, respectively). However, none of the dogs identified as low risk using the distraction index was identified as low risk by either of the NZVA scores. This poor agreement (Kappa value <0) was not improved by using a threshold of >1 in any of the categories used to calculate the NZVA score. The agreement between the two scores was improved by using different thresholds; distraction index >0.5 and an NZVA total score of >9. These thresholds identified a much lower proportion of ‘at-risk’ dogs (5/47 and 10/47, respectively) than the standard thresholds. The Kappa values using the thresholds of 0.5 for the distraction-index and ≥9 for the NZVA scores were moderate for subtotal (0.55; 95% CI=0.16–0.94) and low for total (0.31; 95% CI=−0.02 to 0.63) score. CONCLUSIONS: The low level of agreement between the NZVA total scores and the PennHIP distraction index, particularly when the standard interpretations were used, is of concern as assessing dogs by each method gave disparate results. There was evidence from this study that using the subtotal score moderately improved the level of agreement with distraction index, but only at the higher thresholds. CLINICAL RELEVANCE: The low level of agreement between NZVA and PennHIP results in the same dog precludes them being used interchangeably to guide breeding decisions. The higher heritability of distraction-index measurement in previous studies suggests that it is a better selection tool for breeding dogs when CHD is present within a population. The advantage of a hip-extended ventrodorsal view is its low cost and widespread availability but comparisons between individuals may not be accurate due to the poor sensitivity and the presence of false negatives.