Projected growth of the adult congenital heart disease population in the United States to 2050: an integrative systems modeling approach

CHD mortality

The multiple-cause mortality files for all ages for 1968 to 2010 contain records for
92.4 million deaths by any cause. We identified 288,813 deaths (0.31 %) associated
with a CHD. In 73.4 % of these deaths (212,116, or 0.23 % of all), the CHD was the
underlying cause, i.e., the death was coded as due to a CHD. Between 1968 and 2010,
the cause-specific mortality rate for all ages declined 71 %, from 4.9 to 1.4 per
100,000 person-years (PY) (Fig. 2). The mortality was higher among males (1.5 per 100,000) than among females (1.3
per 100,000). Among zero to 51-week-olds, it declined 69 % whereas among adults aged
20–64 years, the mortality declined 55 %. The percentage of CHD deaths that occur
during the first year of life has declined from 61 % in 1968 to 46 % in 2010.

Fig. 2. Congenital-heart-disease-specific mortality rates as a function of time from 1970
to 2010, stratified by age group. (Markers show National Vital Statistics System data
and lines show model estimates.)

CHD prevalence

The NHIS data from 1997 to 2011 contains recalled CHD status for 180,766 individuals,
from which our model estimates that the birth prevalence of recalled CHD in 2010 for
males was 3.29 per 1,000 (95 % UI 2.8–3.6), and for females was 3.23 per 1,000 (95 %
UI 2.3–3.6) (Fig. 3b).

Fig. 3. a Recalled congenital heart disease prevalence per 1000 as a function of age, stratified
by year of birth. b Recalled prevalence per 1000 at birth (solid red), for 1–4 year olds (dashed green),
and for 5–19 year olds (dotted purple) as a function of time, based on data from the
National Health Interview Survey (circle and square markers)

Prevalence of adult congenital heart disease

Figure 3a shows the prevalence of recalled CHD over time by age group. In 1968, there were
118,000 (95 % UI 72,000–150,000) adults with recalled CHD. By 2010, there was an increase
by a factor of 2.3 (95 % UI 2.2–2.6) to 273,000 (95 % UI 190,000–330,000) adults.
In 2010, there were 134,000 (95 % UI 69,000–160,000) reproductive-age females (age
15–49 years) with recalled CHD.

The estimated number of adults (age 20 to 64 years) with recalled CHD, as well as
the trends in prevalence from 1970 to 2050, are shown in Table 1 and displayed visually in Fig. 4. There will be an estimated 355,000 (95 % UI: 266,000–415,000) adults age 20 to65
years with CHD in 2025 and 510,000 (95 % UI: 400,000–580,000) in 2050. This corresponds
to an estimated 1.47 per thousand adults with CHD in 2010, 1.83 per thousand adults
in 2025, and 2.31 per thousand adults in 2050. The prevalence of adults with recalled
CHD begins to plateau around the year 2050 and growth in the total number of adults
with CHD slows to match the general population growth.

Table 1. Estimated cases and prevalence of ACHD over time

Fig. 4. Estimated number of adults (ages 20 to 64 years) with recalled congenital heart disease
cases (blue solid line) and prevalence of recalled congenital heart disease in adults
(per 1000) (green dotted line), with 95 % uncertainty intervals shaded in grey, as
a function of time, from 1965 to 2060. Previous estimates of ACHD prevalence are marked
with circles for comparison

In 2010, the number of adolescents with recalled CHD who turned 20 years old was 9,800
(95 % UI: 7,700–11,000). This number will increase by a factor of 1.29 to 13,000 (95 %
UI: 10,000–14,000) in 2050. In addition, there will be 170,000 (95 % UI: 100,000–200,000)
reproductive-age females in 2025 and 216,000 (95 % UI: 140,000–250,000) in 2050.

Main findings

We found 273,000 adults had recalled CHD in 2010, which corresponds to 1.47 per thousand
people. This estimate is lower than a recent meta-analysis, which found the adult
CHD population prevalence to be around three per thousand, but they were limited by
the large heterogeneity of the studies and did not include the US 13]. Our integrative systems model confirms that the population of adults with CHD is
increasing; however, we are the first to note that the prevalence will start to plateau
around the year 2050 unless there are significant changes in birth prevalence or mortality.
This has implications for the health care system since it allows us to quantify the
magnitude of this patient population and help with planning for their future healthcare
needs. Previous observational studies of adults with CHD have focused on the severity
distribution, and found that severe cases constitute 3 % and moderate cases 15 % of
the total; none have projected future population trends 4], 13], 28]. Registry data are thus far lacking, and it is difficult to establish registries
that capture adult CHD patients with heterogeneous diseases who may not receive regular
healthcare 29]. Mathematical modeling has been recommended as an important tool for understanding
the burden of this disease 22].

This study used integrative systems modeling to show that the mortality from CHD has
declined in all age groups between 1968 and 2010 with children zero to 51 weeks having
experienced the greatest decline, followed by those 1–4 years, which is consistent
with prior studies that have shown a mortality decreased between 31 and 39 % in this
population 2], 3], 30]. Despite improvements in mortality, these age groups continue to experience the highest
mortality from CHD. Recent implementation of newborn pulse oximetry screening programs
for critical CHD 31], as well as increased perinatal screening using genetic testing and fetal echocardiography,
allows earlier detection of CHD. Earlier diagnosis will continue to decrease the infant
mortality since late diagnosis is associated with worse infant survival 32], 33]. However, we still have much work to do in individuals aged 20 to 64 years old, whose
mortality has remained stable with little improvement over the past 30 years 7]. Our data suggest that the mortality in CHD beyond the neonatal period is shifted
well into adulthood, though still premature compared to the general population.

The care of adult CHD patients requires specialized training, which is being developed
after the 2012 approval of adult CHD as a cardiology subspecialty by the American
Board of Medical Specialties. The current American Heart Association and American
College of Cardiology guidelines recommend that adults with moderate and severe CHD
be seen every 12 to 24 months by a cardiologist with specific CHD expertise at a regional
CHD center; the absence of symptoms is not a reliable indicator of cardiac function
17], 34]–36]. The leading cause of death in the adult CHD population is sudden death (26 %), followed
by progressive heart failure (21 %) and perioperative death (18 %) 34]. Unfortunately, less than half of adolescent patients have adult cardiology follow-up
in a timely manner after they turn 18 years old 19], 37].

This rapidly increasing CHD population in adults, as well as the increasing population
of atherosclerotic heart disease in adults 38], will lead to a substantial increase in health care utilization 39]–41] and increase demand for CHD trained cardiologists. It also requires an estimated
one specialized cardiac center per two million population, which are not yet in place
42]. In addition, CHD patients benefit from multi-disciplinary care team to address their
complex needs given their increased risk for developmental disabilities 16], comorbidities 43], and special considerations for patients desiring pregnancy 44], 45]. Collaborative and multidisciplinary strategies are urgently needed, such as systems
and processes to improve transitions of care from pediatric providers to the adult
health care system 20], 37], 46], improved specialized cardiac centers with cardiologists trained in adult CHD management
42], and technological and medical improvements in care for these patients 42], 47].

Limitations

Our data on CHD prevalence come entirely from the NHIS, a population-based household
survey subject to all the challenges of survey research. Although NHIS has a high
response rate (close to 90 %), it is possible that there is differential non-response
bias, where households with children with moderate to severe CHD are more likely than
average to refuse to be interviewed. We attempted to minimize the bias that this would
introduce by excluding the data from those with children in the zero to 51 week age
group, and sensitivity analysis (Additional file 3) showed that including it reduced the overall prevalence by at most 4 %. CHD is a
group of individually rare diseases and so it is likely unrecalled or unreported when
present (it is also sometimes undiagnosed or diagnosed later in life, making recall
impossible). It is also possible that CHD is reported in cases where it is not present,
due to confusion about the question. However, we believe that recalled CHD is an underestimate
of the overall prevalence of CHD, which reflects the lack of public understanding,
even among those with the condition. As the question only asked for children under
the age of 18, it cannot capture CHD cases diagnosed in adulthood. The recency effect
predicts that this subpopulation would better recall their CHD status than those diagnosed
in childhood, but we suspect that they constitute a small proportion of all ACHD cases.
Despite the large sample size of the NHIS, recalled CHD is rare enough that the data
are quite noisy (Fig. 3b). Our results show that since 1997, there has been a birth prevalence of recalled
CHD of around three per 1,000, which we would like to use as an estimate of moderate
to severe cases of CHD. This is lower than recently published birth prevalence of
four to seven per 1,000 1], 3]–5]. At birth, about half of all CHD are characterized as moderate to severe; in adulthood
moderate CHD accounts for around 38 % and severe around 15 % 13], 17]. Therefore, as the aim of our study was to capture those aware of their moderate
to severe CHD who would survive to adulthood, to determine the current and future
population of adults that will enter the health care system, we feel this reflects
the minimum number of adult CHD patients we would expect to engage in the health system
unless increased public health awareness on the individual and population level are
improved.

Our data on CHD mortality come from an ICD-coded nationwide database, which may have
a low specificity due to miscoding. It may also be nonspecific 48] leading to under-reporting 49] or over-reporting 50]; however, the sensitivity of the system has remained largely the same so the time
trends should still be valid. The mortality rates may be underestimates as the patient
or family and provider need to be aware of the CHD diagnosis in order for it to be
coded on the death certificate. We noted minor variations in the data between ICD
8 and 9 and between 9 and 10 but this did not change the mortality trends. An alternative
line of research has used National Birth Defects Prevention Network (NBDPN) data to
investigate the survival rates for (and birth prevalence of) specific defects 6], 9]–12], 51]. This provides complementary information to our study, but, since it has been more
focused in its cause lists (e.g. considering birth prevalence or survival for hypoplastic
left heart syndrome only 52]), it cannot be incorporated directly into our model. Developing estimates and projections
for more focused cause lists that bring together NBDPN measurements and death certificate
data is an interesting direction for future work. We anticipate that future researcher
will use other approaches and/or refine those used here to further add to our knowledge
of the prevalence of ACHD.

Our finding of decreased infant mortality is consistent with other studies and is
not likely due to a decrease in births, which was adjusted for in our model. Other
possible explanations are also unlikely, but include a decreasing birth prevalence
with an increasing diagnosis of CHD. However, published studies suggest the opposite,
with slightly increased birth prevalence given the increase in the diagnosis with
perinatal screening with fetal echocardiography and neonatal screening with pulse
oximetry 1], 3], 5], 30], 53]. Also, birth prevalence may increase with increased number of adults with CHD having
children with CHD 54]; however, this may be offset by increased termination of pregnancy due to high complex
CHD malformations 55]. Data related to termination of pregnancy are limited and trends are unknown 56]. Thus, our findings of an increased population of adults with recalled CHD due to
the decrease in mortality and increased life expectancy will lead to a growing population
in the future.