I read with interest the report entitled “Radiographic heart size and its rate of increase as tests for the onset of congestive heart failure in Cavalier King Charles Spaniels with mitral valve regurgitation” (Lord et al. JVIM 2011;25:1312-1319). This well-performed study demonstrates that radiographic vertebral heart size (VHS) changes most rapidly in the period of time immediately preceding development of congestive heart failure (CHF). Specifically, in the 8.6 months preceding CHF, the VHS changed by an average of 0.17 vertebra/month as compared to 0.03 vertebra per month during earlier intervals. This finding raises several questions regarding practicality and timing. Based on the reported likelihood ratios, a rate of change of VHS > 0.08 vertebra/month was the best predictor of onset of CHF, and superior to measures of absolute VHS at any time point prior to CHF. Thus, in the 6 to 9 months preceding CHF, an absolute change in VHS of 0.48 to 0.72 vertebrae would signal risk of CHF; however, this change can be less than the reported intraobserver variability (0.6–0.7 vertebrae) of the VHS method.
Furthermore, the study raises the question of when to initiate more frequent radiographic vigilance. Without additional examinations within time frame interval 1 (ie, the 8.6 months preceding development of CHF), detection of an increased rate of change would not occur prior to the onset of CHF (Fig 1B). Indeed, if one were to rely on change in CHF alone, there is nothing in the 12 months prior to time 1 that would alert a clinician to recheck heart size more frequently in the subsequent months. In other words, the increased rate of change is identified one interval too late. CHF has already occurred at the end of the interval of interest, precluding the ability to intervene and potentially change the outcome. In contrast, the data regarding absolute VHS (Fig 1A) indicates an increase in VHS in the second time interval preceding CHF (ie, between times 2 and 1), suggesting that this data could be acted upon prior to the onset of CHF at time 1.
This data raises the question of whether a risk model combining absolute VHS and rate of change of VHS would be superior to a model using either parameter alone. One might suspect that the two variables are sufficiently interrelated such that this would not be the case, but the data presented in Figure 1 of the original article suggests otherwise.