Sound advice for heart murmurs

"A heart murmur! What does this mean? What can be done?" These are questions that any small animal practitioner will have been asked by an owner on many occasions throughout their career. While it might seem that these questions are simply: "Does the puppy have heart disease?" and if it does "What is the treatment?", rephrasing these questions into the language of evidence-based medicine, the questions we find ourselves answering are "What is the risk of the puppy experiencing clinical signs or dying prematurely if we do nothing?" (prognosis question) and (if disease is present) "How is this risk altered by an intervention?" (treatment question). As heart murmurs can be present in young dogs with a variety of congenital heart diseases (Oliviera et al. 2011) and also in dogs without heart disease (Bavegems et al. 2011, Höglund et al. 2011), the answer is rarely a simple one. A good "rule of thumb" is that a low-intensity left basilar murmur in a puppy may be associated with a healthy heart and may disappear by adulthood. For these, we may adopt a "wait and see" policy. However, for some owners such a policy may not be acceptable; many young dogs have murmurs that are not of low intensity, and for some heart diseases earlier detection and treatment may confer a better prognosis (Summerfield et al. 2012). Consequently, we are often in the position of having to determine whether heart disease is present. Echocardiography is the most effective non-invasive tool for diagnosing heart disease. Thoracic radiography is readily accessible for most general practitioners but has been shown to be an inaccurate method for diagnosing congenital heart disease in dogs (Lamb et al. 2001). Absence of heart disease tells us that the puppy has no worse a prognosis than a puppy without a heart murmur. However, if congenital disease is present, this knowledge alone does not give us sufficient information to answer the relevant questions about prognosis and treatment. For this we need to know the type and severity of the heart disease that the puppy has. This information can only be determined by echocardiography. The four most common congenital heart diseases that are consistently reported in prevalence studies in dogs are pulmonic stenosis (PS), aortic stenosis, patent ductus arteriosus and ventricular septal defect (Tidholm 1997, Baumgartner & Glaus 2003, Oliviera et al. 2011). Each condition, when severe, can result in clinical signs and premature death. However, not every dog with one of these conditions will develop clinical signs or die prematurely. For example, dogs with mild aortic stenosis may not experience any decrease in lifespan (O'Grady et al. 1989). Dogs with moderate PS may not have a discernibly worse prognosis than those with mild disease (Francis et al., 2011). We need more information in order to answer the prognosis question. All four conditions are potentially amenable to treatment with an interventional procedure in dogs: balloon valvuloplasty (BV) for PS (Johnson et al. 2004) and aortic stenosis (Meurs et al. 2005); device closure of patent ductus arteriosus (Nguyenba & Tobias 2008) and ventricular septal defect (Bussadori et al. 2007). These procedures have one thing in common: they correct or improve the haemodynamic abnormality that is imposed on the heart by the congenital defect. While it might seem intuitive that a procedure to correct or improve a haemodynamic abnormality should confer a better outcome, this may not be borne out by the results of clinical studies. For example, studies have shown a survival benefit for dogs with patent ductus arteriosus that undergo closure versus those that do not (Van Israël et al. 2003) and for dogs with severe PS (pressure gradient >80 mmHg) that undergo BV versus those that do not (Johnson et al. 2004). However, similar studies evaluating surgical correction (Orton et al. 2000) and BV (Meurs et al. 2005) of aortic stenosis in dogs have failed to show a survival benefit over medical therapy (beta-blockers), while a recent study of dogs with aortic stenosis showed no survival benefit for therapy with beta-blockers over no therapy (Eason et al. 2013). Furthermore, some interventional procedures are associated with perioperative risk, including cardiopulmonary arrest (Margiocco et al. 2008) and death – the published mortality rate for BV has varied from 2.4 to 7.3% (Dennis 2012). We also need more information in order to answer the treatment question. Faced with this situation, how can a veterinarian provide accurate information to the owner of a dog with congenital heart disease about prognosis and offer a recommendation about treatment? The answers to these questions come from clinical studies evaluating the natural history of individual diseases and studies evaluating the effect of therapies. The most robust method to assess the effect of any therapy is a prospective randomised controlled trial (RCT). These are becoming more commonplace in veterinary medicine, particularly for evaluating drug therapies (Summerfield et al. 2012) and even for interventional cardiology (Meurs et al. 2005). RCTs are "the uncontested 'gold standard' for establishing causation between a therapy and outcome" (Nallamothu et al. 2008). When conducted correctly and on a sample of sufficient size, the strength of an RCT is that it can eliminate the most common forms of bias that affect studies of outcome. These are selection bias (differences between control and treatment populations), assessment bias (the effect on a measurement of outcome of the knowledge of whether a subject received treatment) and the effect of suitable subjects being excluded from analysis because of incomplete data or investigator bias. However, RCTs have their limitations. They are resource intensive, are not ideal for assessing complex interactions with outcome, may suffer from limited case recruitment due to ethical or logistical issues and, most importantly, their design tends to result in a highly selected and homogeneous study population (Nallamothu et al. 2008). This final point, which reflects the rigorous experimental design of an RCT is both a strength and a weakness. It is a strength because a homogeneous study population is one way an RCT can eliminate selection bias and generate significant results from smaller samples (reliability). It is a weakness because a homogeneous study population may limit the generalisability of the results of the study to a wider population (external validity). Looked at another way, an RCT is the best single study design for assessing the "efficacy" of a treatment (the beneficial and harmful effects of a treatment in ideal circumstances), but is less suitable for assessing the "effectiveness" of a treatment (the beneficial and harmful effects of a treatment in the circumstances of the real world) (Sackett 2006). A way of assessing the "effectiveness" of a treatment is to conduct an observational study. In observational studies, the populations studied are heterogeneous and set in the real world of the clinic. Their main weakness is that they are subject to all forms of bias. This can be addressed, in part, by studying a sufficiently large population and by using statistical methods that allow not only assessment of associations between treatment and outcome but also evaluation of the interactions and confounding effects of other variables on treatment and outcome. Observational studies are less resource intensive, easier to conduct and may be ethically more acceptable to investigators and participants than RCTs. Another benefit of an observational study over an RCT is that it can provide more reliable data about the natural history of a disease within a real-world population. In this edition of the Journal of Small Animal Practice, Locatelli et al. (2013) report an observational study of 172 dogs with moderate (50–80 mmHg) and severe (>80 mmHg) PS. They were able to evaluate not only the survival times and mortality rates of dogs with moderate and severe PS but also which clinical findings conferred a worse prognosis, and by what magnitude the risk of death was increased by each finding. It is interesting and important that clinical variables that are simple to ascertain (presence of clinical signs, younger age) were just as associated with outcome as complex echocardiographic findings, such as valve morphology and Doppler pressure gradient. It is also the first study to evaluate the effect of BV on survival in dogs with moderate PS (pressure gradient 50–80 mmHg). That no difference in outcome was found for dogs with moderate PS undergoing BV versus dogs that did not undergo BV is as interesting as the mortality rate of the procedure of 0%. Such studies give practising veterinarians the information they need to answer the questions about prognosis and treatment posed by owners of dogs with PS. Now, all we need are similar studies for all congenital heart diseases in dogs and, dare I say it, all diseases in all animals. Simon graduated from the Royal Veterinary College (RVC) in 2000. He returned to the RVC in 2005 as a resident, then as a lecturer in cardiology from 2008 to 2011. He moved to North Downs Specialist Referrals in 2011 and joined the University of Pennsylvania in 2013. He is cardiology associate editor for the Journal of Small Animal Practice and chair of the cardiology examination committee for the ECVIM-CA.