Outcome in 55 dogs with pulmonic stenosis that did not undergo balloon valvuloplasty or surgery

Authors

  • A. J. Francis,

    1. Royal (Dick) School of Veterinary Studies, The Roslin Institute, Division of Clinical Veterinary Studies, The University of Edinburgh, Hospital for Small Animals, Easter Bush Veterinary Centre, Roslin, Midlothian EH25 9RG
    Search for more papers by this author
  • M. J. S. Johnson,

    1. Veterinary Cardiorespiratory Centre, Thera House, Waverley Road, Kenilworth, Warks CV8 1JL
    Search for more papers by this author
  • G. C. Culshaw,

    1. Royal (Dick) School of Veterinary Studies, The Roslin Institute, Division of Clinical Veterinary Studies, The University of Edinburgh, Hospital for Small Animals, Easter Bush Veterinary Centre, Roslin, Midlothian EH25 9RG
    Search for more papers by this author
  • B. M. Corcoran,

    1. Royal (Dick) School of Veterinary Studies, The Roslin Institute, Division of Clinical Veterinary Studies, The University of Edinburgh, Hospital for Small Animals, Easter Bush Veterinary Centre, Roslin, Midlothian EH25 9RG
    Search for more papers by this author
  • M. W. S. Martin,

    1. Veterinary Cardiorespiratory Centre, Thera House, Waverley Road, Kenilworth, Warks CV8 1JL
    Search for more papers by this author
  • A. T. French

    1. Royal (Dick) School of Veterinary Studies, The Roslin Institute, Division of Clinical Veterinary Studies, The University of Edinburgh, Hospital for Small Animals, Easter Bush Veterinary Centre, Roslin, Midlothian EH25 9RG
    Search for more papers by this author

Abstract

Objective: To determine the outcome, independent predictors of cardiac death, and the Doppler-derived pressure gradient cut-off for predicting cardiac death in dogs with pulmonic stenosis, with or without tricuspid regurgitation, that do not undergo balloon valvuloplasty or valve surgery.

Methods: Review of medical records of two UK referral centres between July 1997 and October 2008 for all cases of pulmonic stenosis that had no balloon valvuloplasty or valve surgery. Inclusion criteria included a diagnosis of pulmonic stenosis; spectral Doppler pulmonic velocity greater than 1·6 m/s; characteristic valve leaflet morphological abnormalities. Exclusion criteria included concurrent significant cardiac defects, including tricuspid dysplasia. Dogs with tricuspid regurgitation were included. Dogs were classified according to Doppler-derived pressure gradients into mild, moderate or severe pulmonic stenosis categories.

Results: Presence of tricuspid regurgitation and severe stenosis were independent predictors of cardiac death. A pulmonic pressure gradient of more than 60 mmHg was associated with 86% sensitivity, and 71% specificity of predicting cardiac death.

Clinical Significance: There is an increased probability of cardiac death in those cases which have a pulmonary pressure gradient greater than 60 mmHg and tricuspid regurgitation, though the effect of severity of tricuspid regurgitation on outcome was not measurable because of small sample sizes. These animals might benefit from intervention.

Introduction

Pulmonic stenosis (PS) accounts for between 11% and 18% of cases of congenital cardiac disease (Bonagura and others 2009) and is one of the three most common congential cardiac defects (Tidholm 1997, Ettinger and Feldman 2005). Certain breeds appear over-represented, including some small terrier breeds, the English bulldog, boxer, beagle, mastiff and Samoyed (Fox and others 1999, Bussadori and others 2001). A slight male predilection has been shown in some studies (Tidholm 1997, Bussadori and others 2001).

The stenosis can be subvalvular, valvular or supravalvular in dogs, but congenital valvular stenosis is the most commonly recognised form (Ristic and others 2001).

The severity of stenosis can be determined by several methods (Griffiths and others 2006). The indirect, continuous wave Doppler (CWD) derived peak gradient system appears to be most often utilised: mild stenosis 10 to 49 mmHg, moderate stenosis 50 to 80 mmHg, and severe stenosis more than 80 mmHg (Kittleson and Kienle 1998, Bonagura and others 2009, Luis Fuentes and others 2010), or greater than 100 mmHg (Fox and others 1999). This system appears to have been extrapolated from human data (Kittleson and Kienle 1998, Fox and others 1999). Various studies investigating success rates of balloon valvuloplasty (BV) or valvulotomy in severely affected dogs have been reported (Fingland and others 1986, Martin and others 1992, Bussadori and others 2001, Johnson and Martin 2004). However, the long-term outcome of dogs with PS that did not undergo BV or surgery has only been evaluated in one previous study (Johnson and others 2004).

This paper presents long-term outcome data from dogs with mild, moderate and severe PS, as classified above, and documents variables likely to affect the risk of experiencing cardiac death. It also seeks to determine the Doppler-derived pressure gradient (PG) that would be an appropriate cut-off in predicting cardiac death.

Materials and methods

The medical records of two UK referral centres were reviewed between July 1997 and October 2008 for all cases of PS that had no BV or valve surgery (VS). The major inclusion criterion for the study was a diagnosis of PS, with a PG of more than 10 mmHg (Luis Fuentes and others 2010). Valve morphology was assessed for evidence of valve thickening, immobility, commisural fusion and annular hypoplasia, as an integral part of the diagnosis of PS, though the type (A or B) of lesion was not recorded in all cases.

Exclusion criteria were other concurrent cardiac defects causing clinical signs or evidence of cardiomegaly, but dogs with tricuspid valve regurgitation (TR) or defects considered clinically insignificant were included. Animals with severe tricuspid valve abnormalities, such as Ebstein anomaly, or 2-D echocardiographic evidence of structural tricuspid valvular apparatus abnormalities consistent with tricuspid dysplasia were excluded. A subjective differentiation between TR secondary to annular dilatation and TR secondary to tricuspid valve dysplasia (TVD) was made.

Dogs were classified into mild, moderate or severe categories according to the PG: mild stenosis 10 to 49 mmHg, moderate stenosis 50 to 80 mmHg, and severe stenosis more than 80 mmHg (Bonagura and others 2009, Luis Fuentes and others 2010).

Echocardiography was performed from both right and left sides, in a standard manner, with the pulmonic velocity assessed from the right parasternal short axis and left cranial parasternal views in all cases (Bussadori and others 2000). Optimal spectral velocity tracings were obtained, and the single maximum pulmonic velocity was converted into a PG for classification, using the modified Bernoulli equation.

TR, when present, was subjectively noted as mild, moderate or severe, based on the area of the colour-flow Doppler map (Fox and others 1999). Right ventricular hypertrophy (RVH), where present, was noted and classified as mild, moderate or severe, using the criteria of Johnson and others (2004). Right atrial size was subjectively evaluated as normal or enlarged.

In cases with a PG more than 80 mmHg, BV was recommended as the initial treatment of choice. Reasons for not performing surgery or BV included financial concerns, owner’s preference, and failure to return for follow-up. Any medications prescribed were recorded.

Follow-up consisted of annual clinical re-evaluation and echocardiography wherever possible. For dogs that were not re-presented, follow-up was performed via telephone conversations with the owners and the referring veterinary surgeons.

Final follow-up was performed via telephone conversations with the owners and referring veterinary surgeons during the week ending on the 15th of May 2009. Owners were asked whether their pet was alive or dead. If alive, then the presence or absence of clinical signs attributable to right-sided congestive heart failure, i.e. ascites, was ascertained. If dead, the date and circumstances of death were obtained. All deaths were assumed to be cardiac related, including those animals that had died suddenly, unless there was a clearly unambiguous non-cardiac cause of death; for example, neoplasia, status epilepticus, severe gastrointestinal disease. If owners were not contactable or traceable via their referring veterinary surgeon after the initial examination, then they were recorded as lost to follow-up.

Statistics

Data were tabulated and organised in Microsoft Excel® before being exported into the commercial statistics program R (v2.8.1 © The R Foundation for Statistical Computing 2008) for statistical analysis. For continuous variables, medians and interquartile ranges were calculated. Most data were binomial, so descriptive statistics consisted of proportions. Univariate logistic regression using the generalised linear model for proportional data was performed, assuming a binomial distribution. Results were expressed as odds ratios with 95% confidence intervals (CI). Statistical significance was assessed using Wald’s test, with a P value of < 0·05.

Predictor variables identified by univariate logistic regression as having an effect on time to cardiac death, defined as having a P value < 0·1, were included in a multivariate logistic regression model. Variables were tested for interaction, but no statistically significant interactions were found. Multivariate logistic regression was performed using the generalised linear model for proportional data, and the final model was obtained via stepwise deletion. A receiver operating characteristic (ROC) curve of sensitivity versus 1 – specificity with 95% CI was plotted to ascertain the sensitivity and specificity of using the Doppler-derived pulmonic PG as a predictor of cardiac death, and the area under the curve (AUC) was calculated.

Results

A total of 76 cases met the inclusion criteria, but 21 were excluded, leaving data from a total of 55 dogs available for analysis. Reasons for rejection included no follow-up in 6 cases and concurrent clinically significant disease, including tricuspid valve dysplasia, in 15 cases.

All dogs had been presented for the evaluation of either a heart murmur, 54 dogs (98%), or clinical signs including ascites, exercise intolerance, lethargy or collapse, 14 dogs (25%). The one dog in which a murmur was not noted on auscultation was an English bull terrier that had been presented for mild exercise intolerance. The lack of murmur was attributed to conformational difficulties with auscultation. The median murmur grade was 4/6, with a precordial thrill palpable in 16 dogs (29%). The median heart rate was 120 beats per minute.

Thirty-one animals (56%) underwent electrocardiography (ECG), of which four animals had a documented arrhythmia; ventricular premature complexes (VPCs) in two dogs and inappropriate sinus tachycardia in two dogs. Twenty-seven dogs (49%) had thoracic radiography performed.

All animals underwent echocardiography; RVH was noted in 35 dogs (64%), with right atrial enlargement in 6 dogs (11%). Tricuspid regurgitation was noted in 19 dogs (35%). The underlying cause of regurgitation was considered unknown in 13 dogs, and due to annular dilatation in 6 dogs.

Twenty-three different breeds, including cross-breeds, were represented (Table 1), with the cocker spaniel as the modal breed. There were 26 females (47%), out of which 5 were neutered, and 29 males, (53%), out of which 4 were neutered.

Table 1. Frequency of breeds affected by pulmonic stenosis in this study
BreedNumber
Cocker spaniel9
Labrador retriever6
Cross breed4
West Highland white terrier4
English bulldog3
English bull terrier3
German shepherd dog3
Golden retriever2
Border collie2
Border terrier2
Boxer2
Bull mastiff2
Cavalier King Charles spaniel2
English springer spaniel2
Pyrenean mountain dog1
Basset hound, beagle, Chihuahua, Flat-coated retriever, Jack Russell terrier, Lhasa apso, miniature bull terrier, shih tzu.1 of each breed

Forty dogs (72%) were under 20 months of age at first presentation (median age 11·0 months, interquartile range 4 to 24 months), but a minority were much older, with one dog presented at 9 years of age (Fig 1).

Figure 1.

Histogram illustrating the frequency distribution of the age of the dogs at the time of first examination

On initial evaluation, the owner was aware of the presence of clinical signs in 14 dogs (25%). Exercise intolerance and syncope were the most frequently reported clinical signs, seen in 7 dogs (13%) and 6 dogs (11%), respectively, while lethargy was reported in 4 dogs (7%). Within groups, 7 out of 26 mildly affected dogs (26%), 3 out of 21 moderately affected dogs (14%), and 4 out of 8 severely affected dogs (50%) had clinical signs at initial presentation. Only one dog (2%) had ascites on initial presentation, and was therefore classified as having right-sided congestive heart failure. This dog had severe PS, with moderate tricuspid regurgitation and significant right atrial enlargement.

Twenty-six dogs (47%) were classified as having mild PS, 21 dogs (38%) had moderate PS and 8 dogs (15%) severe PS. Median pulmonic PGs in each group were 29, 60 and 125 mmHg, respectively (Fig 2).

Figure 2.

Box and whisker plots comparing the range of pulmonic pressure gradients measured on first examination between groups. Thick horizontal line median value: box denotes interquartile range; broken line represents data range

Medications were prescribed in five dogs on initial examination, and included atenolol (Atenolol; Chatfield Pharmaceuticals), benazepril (Fortekor; Novartis Animal Health) and furosemide (Frusecare; Animalcare) (Table 2).

Table 2. Number of dogs prescribed particular medications or medication combinations at the initial evaluation, depending on their severity of stenosis
DrugSeverity of stenosisTotal number of dogs given particular medications
Mild stenosisModerate stenosisSevere stenosis
Atenolol0224
Benazepril, furosemide0011

Of the 55 dogs originally enrolled, 34 dogs (62%) were alive at the end of the study, 12 dogs (22%) had died and 9 dogs (16%) were unavailable for final follow-up. The median follow-up time was 1096 days (interquartile range 648 to 1926 days).

Seven dogs (13%) had a cardiac cause of death, with four dogs euthanased because of right-sided heart failure, and sudden death recorded in three dogs. One dog was euthanased due to left-sided congestive heart failure and was not classified as having suffered cardiac death. Four dogs had a non-cardiac cause of death, of which two dogs were euthanased due to neoplasia, one dog was euthanased due to severe GI disease and one dog died at home after prolonged seizure activity. The cardiac death rate in mildly affected dogs was 4% (one dog), in moderately affected dogs 10% (two dogs), and in severely affected dogs 50% (four dogs) (Table 3). The dog with mild PS that died had pulmonic PGs of between 46 and 48 mmHg, evidence of thickened pulmonic valves, significant tricuspid regurgitation but no morphological evidence of tricuspid dysplasia, and right atrial enlargement.

Table 3. Outcome data of dogs, based on their severity of stenosis at initial presentation
Stenosis severityNumber of dogs on day 0Outcome at final follow-up
AliveCardiac deathLeft-sided heart failureNon-cardiac death
EuthanasiaSudden cardiac death
Mild stenosis26211013
Moderate stenosis21181101
Severe stenosis842200
Total55434314

Univariate logistic regression analysis was performed to assess for predictors of cardiac death due to right-sided heart failure or sudden cardiac death, and odds ratios were calculated (Table 4).

Table 4. Results of the univariate logistic regression analysis of outcome for each variable, the odds ratios, 95% confidence intervals, and P value indicating statistical significance as assessed using Wald’s test
Risk factorsOdds ratio95% Confidence intervalsP value
  1. * Reached statistical significance with a P value of less than 0·05

Signalment
 Age at first examination0·990·95 – 1·040·80
 Male compared to female1·230·25 – 6·080·80
 Neutered – compared with entire2·340·38 – 14·540·36
 Breed
  Small (reference level)
  Medium breed2·100·20 – 22·330·54
  Large breed1·080·06 – 19·050·96
  Giant breed281·21 – 648·810·038*
 Clinical signs on initial evaluation
 Presence of clinical signs2·520·49 – 13·020·269
 Lethargy35·252·94 – 422·090·005*
 Exercise intolerance8·251·35 – 50·580·023*
 Syncope0·950·21 – 15·750·98
 Presence of clinical signs excluding syncope4·390·80 – 24·00·088
Physical examination findings
 Precordial thrill2·020·40 – 10·270·397
 Arrhythmia2·800·44 – 17·800·275
 Heart rate0·990·97 – 1·020·918
 Murmur grade1·860·71 – 4·830·205
Electrocardiography (ECG)
 Arrhythmia on ECG5·250·56 – 48·950·145
 Right axis deviation2·550·39 – 16·550·328
Echocardiography
 Group – based on peak pulmonary pressure gradient
  Mild (reference level)
  Moderate2·630·22 – 31·220·443
  Severe252·20 – 284·610·009*
 Combined groups
  Severe compared with mild and moderate combined14·672·39 – 89·930·004*
 Tricuspid regurgitation (TR)16·151·77 – 147·350·014*
 TR severity
  Mild (reference level)
  Moderate5·500·23 – 128·970·29
  Severe16·51·09 – 250·180·043*
 Combined TR severity
  Severe compared to non-severe120·9 – 160·40·06
 Right ventricular concentric hypertrophy
Severe compared to non-severe6·451·11 – 37·50·038*
 Pulmonic insufficiency (PI)0·370·04 – 3·320·372
 Increased right atrial size11·251·68 – 71·150·012*
Radiography
 Right atrial enlargement1·970·08 – 45·650·738
 Pulmonary arterial bulge4·800·46 – 50·160·190

Overall, dogs showing any clinical signs at first presentation did not have a statistically significantly greater risk of suffering cardiac death (P=0·269). However, dogs presenting with exercise intolerance had an eight-fold greater risk of suffering cardiac death, odds ratio (OR) 8·25 and 95% confidence interval (95%CI) 1·35 to 50·58; those animals with lethargy had a 35-fold increase in risk, OR 35·25 (95%CI: 2·94 to 422·09). Because syncope was a very poor predictor of cardiac death, the presence of clinical signs, excluding syncope, was evaluated as a covariate, but this failed to result in a statistically significant association (P=0·088).

Dogs with TR were at 16 times greater risk of cardiac death, OR 16·15 (95%CI: 1·77 to 147·35). Right atrial enlargement was associated with an 11-fold increase in risk of cardiac death, compared with those dogs without right atrial enlargement, OR 11·25 (95%CI: 1·68 to 71·15). The presence of severe right ventricular concentric hypertrophy (RVCH) was associated with a six-fold increase in risk of cardiac death, OR 6·45 (95%CI: 1·11 to 37·50).

When the severity of the PS (mild, moderate or severe) was considered as an individual covariate, there was a 25-fold increase in risk of cardiac death in severely affected dogs, OR 25 (95%CI: 2·20 to 284·61).

The covariates included in the initial multivariate model were clinical signs at first presentation, exercise intolerance, an ECG arrhythmia, presence or absence of severe stenosis, and tricuspid regurgitation; no interactions were noted between the different variables. The final model had the remaining two factors: presence of severe stenosis and presence of tricuspid regurgitation (Table 5).

Table 5. Final multivariate logistic regression, generalised linear model predicting cardiac death
Risk factorsCrude OR95% CIAdjusted OR95% CIP value
  1. CI Confidence interval

  2. The crude odds ratio (crude OR) is the odds ratio of the variable obtained via univariate logistic regression. The adjusted odds ratio (adjusted OR) represents the change in the odds ratio when the variables are combined. The P value is the statistical significance as assessed by Wald’s test

  3. * Reached statistical significance with a P value of less than 0·05

Severe stenosis Y versus N14·672·39 – 89·9311·251·44 – 87·640·021*
Tricuspid regurgitation Y versus N16·151·77 – 147·3513·041·26 – 135·210·031*

Finally, the ROC curve estimating what PG at initial presentation best discriminated for cardiac death (Fig 3) revealed that a PG of 60 mmHg maximised the sensitivity (86%) and specificity (71%) of predicting cardiac death using the PG at initial presentation (Table 6).

Figure 3.

Receiver operating characteristic (ROC) curve. Predicting cardiac death by measurement of pulmonic pressure gradient at first examination. AUC=area under curve

Table 6. Best cut-off estimations of the sensitivity and specificity of the pulmonic pressure gradient (PPG) on initial evaluation, as a predictor of cardiac death, with 95% confidence intervals
PPGSensitivitySensitivity 95% CISpecificitySpecificity 95% CI
  1. CI Confidence interval

1200·4290·158 – 0·7500·9580·860 – 0·989
640·7140·359 – 0·9180·7290·590 – 0·834
600·8570·487 – 0·9740·7080·568 – 0·818

Discussion

This is the first study to present long-term follow-up data for all dogs presenting with PS, regardless of the severity of stenosis. Other studies have documented clinical outcomes and long-term follow-up after BV (Bussadori and others 2001, Johnson and Martin 2004, Johnson and others 2004), pericardial patch grafting, and closed valvulotomy (Fingland and others 1986), comparing outcome in severely affected animals that underwent pulmonic valvular dilatation with those severely affected animals that did not.

In this series, 53% of the dogs were male and 47% female, a sex distribution in agreement with a number of studies (Fingland and others 1986, Tidholm 1997, Bussadori and others 2001).

The most commonly presented breeds in this study were the cocker spaniel and Labrador retriever. Though the modal breed is quite variable between studies, and likely represents variation in the local dog population, the results presented in this study are in agreement with the most frequently reported breeds in other studies (Tidholm 1997, Ristic and others 2001, Bonagura and others 2009).

The majority of dogs, 72%, were less than 20 months of age on initial presentation, and the median age at first presentation, 11 months, is in agreement with another study (Johnson and Martin 2004).

The percentage of dogs with clinical signs at first examination, 25%, is less than that reported in other studies (Bussadori and others 2001, Ristic and others 2001, Johnson and others 2004). This may be because other studies have evaluated dogs with severe stenosis, and these animals may be more likely to present with clinical signs. In our study, 50% of severely affected dogs showed clinical signs. Syncope and exercise intolerance were the most commonly reported clinical signs, in agreement with the literature (Bussadori and others 2001, Ristic and others 2001).

No previous studies have documented the cardiac death rate of mild, moderate and severely affected animals separately. In our study, 50% of severely affected dogs, 10% of moderately affected dogs and 4% of mildly affected dogs suffered cardiac death. The cardiac death rate of severely affected dogs that did not undergo BV or surgery has only been reported in one previous study; 34% of dogs died suddenly or were euthanased due to refractory right-sided heart failure (Johnson and others 2004). The difference between the values presented in this study and those reported by Johnson and others (2004) could represent variation in the follow-up periods between the two studies; the median follow-up period in this study was 1096 days, and in the Johnson study long-term follow-up was defined as follow-up of 6 months or more.

Univariate logistic regression revealed a number of covariates that individually predicted an increased risk of cardiac death, though a number of these covariates had odds ratios with large 95% CI. This is likely to be because of the small numbers of dogs presenting with specific abnormalities, and consequently, covariates with data points for all dogs enrolled in the study were evaluated wherever possible, so as to minimise the variation in CI and minimise the effect a single dog might have had on the data set.

The multivariate analysis showed the presence of TR to be an independent negative prognostic indicator, when the effect of pulmonic PG on risk of cardiac death was kept fixed. This might be explained by the theoretical effect of the presence of significant TR decreasing flow across the pulmonic valve, leading to a decrease in the peak pulmonic velocity measured. A similar effect has been previously documented in the diagnosis of severity of subaortic stenosis (Bonagura 2001).

The scope of the multivariate logistic regression analysis was limited by the small numbers of dogs in some groups, because multivariate regression is even more dependent on large covariate sample sizes than univariate analysis. Ideally, larger prospective studies would be required to further evaluate and corroborate these findings.

In human medicine severe TR is usually noted in adults with uncorrected congenital PS, secondary to tricuspid annular dilatation (Fawzy and others 2007). Tricuspid valve incompetence is recognised as adversely affecting prognosis (Brickner and others 2000), though it tends to resolve or decrease significantly after BV, even if severe (Fawzy and others 2007).

Future studies therefore need to reclassify the severity of stenosis according to the presence or absence of TR, as well as using pulmonic PGs.

Echocardiography to assess severity of PS can be regarded as a screening test to ascertain the risk of cardiac death. With any screening test, sensitivity should be maximised, though this does decrease the specificity (Bland 1995). It is arguably better to perform BV in a few dogs unnecessarily, than to fail to perform BV in an animal that requires it. A PG required to achieve 100% sensitivity would likely result in an unacceptably low specificity, so both sensitivity and specificity were maximised resulting in a cut-off PG of 60 mmHg. The CI were again fairly variable, and further studies would be required, using a larger number of cases.

There were a number of study limitations that restricted the number and scope of the conclusions that could be drawn from the data. More detailed evaluation of the structural causes of the documented tricuspid regurgitation in individual cases may have facilitated interpretation of our data and this would be beneficial in any future prospective study. However, the definitive echocardiographic diagnosis of the cause of TR is difficult, unless obvious structural tricuspid valve apparatus abnormalities are noted, and these dogs were excluded. The retrospective nature of the study meant that confounding variables were poorly controlled between the mild, moderate and severely affected groups. Beta blockers were prescribed in 7% of cases on initial evaluation. The decision to medicate was at the discretion of the attending clinician, and may have had an effect on outcome. No studies have evaluated long-term effects of early use of beta-blockade in animals with PS (Bonagura and others 2009), so there is no indication of how beta-blocker medication might have affected outcome.

The numbers of animals in each group were also unequal, with far fewer animals in the severe group, because most animals with severe stenosis undergo BV. Whether the severe group sample was representative of the population of severely affected dogs in general is unknown. A prospective study would also have allowed standardisation of the tests performed in each case on initial evaluation. One of the major problems with the multivariate analysis was the low numbers of animals that exhibited a particular clinical sign or underwent a particular diagnostic test. As a result of the relatively low case numbers, it was not possible to make distinction between various grades of TR severity as risk factors in the multivariate analysis. Greater numbers of dogs overall would have made a significant difference in this respect.

In conclusion, the scope of this study is limited by the relatively small sample sizes, though a prospective study with larger numbers of dogs would be difficult to perform, because most severely affected dogs now undergo BV, and it would be ethically unacceptable not to offer BV in animals with severe stenosis. The 80 mmHg cut-off value for recommending intervention is a sensible starting point, but it is possible that a number of animals with PG less than 80 mmHg would benefit from BV or VS. The multivariate analysis showed the presence of TR to be an independent negative prognostic indicator, when the effect of pulmonic PG on risk of cardiac death was kept fixed. Therefore, it is likely that dogs with TR, and a systolic PG across the pulmonic valve greater than 60 mmHg may benefit from BV or VS. Future follow-up studies assessing outcome related to pulmonic stenosis severity should take TR into account.

Acknowledgements

Thanks to Dr D.J. Shaw, BSc, PhD, for statistical advice. Thanks are also due to the owners and referring veterinary surgeons of the animals in this study, without whom this study would not have been possible.

Conflict of interest

A. J. Francis currently works for Boehringer Ingelheim. The company was not involved in any aspect of the preparation of this manuscript. None of the authors of this article has a financial or personal relationship with other people or organisations that could inappropriately influence or bias the content of the paper.

Ancillary