The Epidemiology of Persons Living with Fontan in 2020 and Projections for 2030: Development of an Epidemiology Model Providing Multinational Estimates

Approximately 1 in 10,000 infants are born with a univentricular heart [1] and are unlikely to survive into adulthood without the palliative Fontan procedure [1, 2]. In the modern era, the Fontan operation is the final operation of a staged surgical process (at least two major cardiovascular surgeries) conducted to achieve a Fontan circulation. Norwood is performed on infants in the first few weeks of life, followed by Glenn surgery at the age of around 4–6 months and, finally, the Fontan procedure is commonly performed in infants between the ages of 2 and 4 years [3‐7]. As a result of the surgery series, both superior and inferior venae cavae are rewired so that venous return is routed directly to the pulmonary artery, bypassing the heart, thus preventing oxygenated and deoxygenated blood from mixing in the single ventricle [8].

Although the principles of Fontan surgery have remained unchanged over the past few decades, the Fontan technique has been modified since the original procedure. Thus, there are three types of Fontan procedure: atriopulmonary connection (AP), lateral tunnel total cavopulmonary connection (LT-TCPC), and extracardiac total cavopulmonary connection (EC-TCPC). These procedures were first described in 1971 [9], 1983 [10, 11] and 1988 [11, 12], respectively. Over time, TCPC procedures have superseded the original AP surgery [13], becoming more common in the 1990s [14, 15], due to the improved survival rates with TCPC procedures [13, 16‐19]. In the early surgical era (1973–1990), the 10-year survival rate was 69% [19], improving to 89% by 1991–2000 and reaching 95% by 2001 onwards [19]. There is a surgical preference for EC-TCPC in several clinics [20, 21], based on studies indicating a lower incidence of adverse events (such as arrhythmias) following surgery [19‐22]. However, nearly half of all Fontan survivors experience major adverse events in the years following the procedure [23]. Further long-term outcome data would be valuable to fully understand the impact of the latest procedure modifications [13, 20]. Importantly, robust data are lacking on the current epidemiology of persons living with Fontan.

This study aimed to develop an epidemiologic model to estimate the prevalence of persons living with Fontan in 11 countries (Australia, New Zealand, Sweden, Switzerland, Netherlands, Germany, Italy, Spain, France, UK and USA). The prevalence was estimated for 2020 and extrapolated through to 2030, with stratification of prevalence estimates by procedure type (AP, LT-TCPC and EC-TCPC) and age group (< 12 years, 12–17 years and ≥ 18 years old).

This epidemiology model was created to fill a gap in knowledge on the epidemiology of Fontan. Estimates for the numbers of children born with a single ventricle heart are available in the literature, but it is not possible to use these as epidemiology estimates, as not all patients will undergo Fontan surgery or survive the Fontan procedures. Most studies published to date are small and/or based on a single centre and therefore cannot be relied upon for robust epidemiology estimates [20, 30, 31]. Our model estimated the prevalence of persons living with Fontan across the 11 included countries to be 66 ppm in 2020, rising to 79 ppm in 2030. In addition, our model estimated that the Fontan population would become increasingly older (55% adults in 2020, rising to 64% in 2030) and that the proportion of persons living with an EC-TCPC procedure type would also increase (74–83%). This model has collated the largest set of multinational data on Fontan epidemiology, and represents the only set of multi-continental data, to our knowledge. As such, this model provides novel insights into the epidemiology of persons living with Fontan.

Our model suggests that the number of persons living with Fontan is likely to continue to grow. The projected growth in our model was driven by increased survival due to TCPC procedures, particularly with EC-TCPC, compared with earlier forms of Fontan [17‐19] and the projected growth in general population size (with the exception of Germany, where population size is in decline). In turn, the model also provided evidence of an ageing Fontan population, as more patients survive into adulthood. This is consistent with another study of Fontan epidemiology: Schilling et al. [14] estimated the prevalence of persons living with Fontan in Australia and New Zealand to be 45 ppm in 2014, rising to 58 ppm in 2025 and 72 ppm in 2045, with the average age of the Fontan population increasing from 18 to 31 years over the same time period. They projected prevalence probabilistically using a microsimulation method and robust data from the long-term Australia and New Zealand registry. Their model also considered more factors relating to survival such as gender and morphology (hypoplastic left-heart syndrome versus other) compared with ours. Our model supports these findings of a growing, ageing Fontan population and suggests this trend applies globally. Therefore, the healthcare support needs of this rare population could increase in future.

Our model is based on survival estimates taken from the literature, and the results of our literature search further demonstrate the scarcity of data available on the long-term survival with Fontan surgery. This is particularly true for the EC-TCPC procedure. Thirty-year survival rates were not available in the literature for EC-TCPC, due to the relatively recent availability of this procedure. It is reassuring that the survival curves generated and used in this model predicted a 30-year survival rate for EC-TCPC (80%) that is in line with the expectations for 30-year survival with EC-TCPC noted in the Scientific Statement from the American Heart Association [4]. However, further research is needed to confirm if the long-term survival rates with EC-TCPC are accurate.

This study has several limitations. The average number of annual Fontan procedures for the 2010–2020 period was based on the five most recent years with data available. Therefore, projections both forwards and backwards in time were based on assumptions and potential fluctuations in trends may not have been adequately captured. There could be fluctuation in the incidence of pregnancy terminations, for example, and improvements in the biventricular repair technique [32‐35] may lead to some decline in the number of Fontan surgeries conducted in future. However, we have assumed a steady rate of Fontan and these potential trends would therefore not be captured. It is also unlikely that the databases used to extract Fontan procedure codes have captured all clinically relevant cases of Fontan. In addition, our model did not account for further interventions, such as conversion of Fontan circulation (AP to TCPC), Fontan takedown or heart transplantation, or for the potential for differences in model parameters to be driven by the type of congenital heart defect. It is also important to highlight that the model cannot be used to provide country-specific estimates, as there were not enough country-specific data to populate the model. Instead, the prevalence estimates are ‘global’ across the 11 included countries.

Our model’s key strength is that it involves collection of the strongest existing evidence of Fontan epidemiology and therefore represents the largest set of data available. Another strength is that the model inputs and outputs were cross-checked with other sources at several stages. For example, the procedure counts extracted from the US COMPILE database were relatively consistent with those reported from the NIS by Akintoye et al. [25], suggesting that the retrieved data are reliable. Moreover, the model’s prevalence outputs were cross-checked with other estimates available in the literature, finding only slight difference between our estimates and those of Schilling et al. [14], Iyengar et al. [28] (both on the Australia and New Zealand registry) and Dahlqvist et al. [29] (Sweden). However, the similarity between our models estimates and those reported by Schilling et al., and Iyengar et al., are expected, given that our model used this registry as the source of the Australia-New Zealand procedure counts and as the basis for creating the uptake curve. Nevertheless, it is reassuring that there are no large differences between our model’s estimates and published estimates of Fontan prevalence in Sweden, Australia and New Zealand.

Our epidemiologic model collates existing evidence from 11 countries across Europe, USA, Australia and New Zealand and provides a novel overview of the epidemiology of persons living with Fontan. Our model indicates that there is a growing and ageing population of persons living with Fontan, with prevalence predicted to increase from 66 to 79 ppm between 2020 and 2030, and the proportion of adults rising from 55 to 64% over the same period. In our model, these changes were driven by the better survival rates with TCPC versus AP, the increasing preference for EC-TCPC and the predicted growth in general population size.