Background
Coarctation of the aorta (CoA) is difficult to diagnose by fetal echocardiogram (F-Echo), often requiring multiple F-Echos during gestation and neonatal echocardiograms (N-Echos) after birth. Furthermore, CoA is the most common ductal-dependent lesion missed on routine physical exam.
Objectives
We sought to determine the most cost-effective diagnostic approach in caring for infants in whom an initial F-Echo is concerning for CoA.
Methods
Four paradigms for management after initial F-Echo could not rule out CoA were compared, with a single paradigm involving additional F-Echos: (1) multiple F-Echos for diagnostic clarity and performance of N-Echo on neonates with remaining high suspicion for CoA on F-Echos (prenatal-multiple), (2) no further F-Echo and performance of N-Echo on neonates with high suspicion for CoA on initial F-Echo (postnatal-selective), (3) no further F-Echo and performance of N-Echo on all neonates (postnatal-all), and (4) no further F-Echo or N-Echo with reliance on routine physical exam to identify afflicted infants (postnatal-none). Decision analysis models were constructed. Probabilities dictating clinical course and costs were calculated using our institution’s study population. The utility-state values were derived from existing literature. The measure of effectiveness was quality-adjusted life years. To represent societal perspectives, cost was defined as hospital reimbursement payments.
Results
From 2007 to 2014 at our institution, 92 patients were diagnosed with CoA and met the inclusion criteria for this study. These patients presented to care either through prenatal diagnosis ( n = 31), postnatal examination findings while clinically well ( n = 41), or after clinical deterioration in extremis ( n = 20), with one patient subsequently dying. Presenting in extremis was associated with a 20% increase in the cost of their subsequent care and with a 51% increase in length of hospital stay. Postnatal-none was the least effective paradigm but also the least costly, thus forming the baseline model. Of the three other diagnostic approaches modeled, Postnatal-all was the cost-effective paradigm, maximizing utility due to avoidance of high-cost/low-utility disease states such as presentation in extremis and death. Prenatal-multiple was the next most effective but was also the most expensive.
Conclusions
Echocardiography is the screening gold standard in avoiding the devastating clinical manifestations of a missed CoA. When a diagnosis of CoA cannot be ruled out on initial F-Echo, the most cost-effective approach is performance of N-Echo on all neonates with no further prenatal evaluation.
Highlights
- •
Prenatal echocardiographic concern for coarctation of the aorta is a frequent diagnostic challenge for pediatric cardiologists.
- •
Faced with diagnostic uncertainty, providers often repeat fetal echocardiograms throughout gestation, yet doing so often does not add diagnostic clarity.
- •
This study illustrates that the optimally cost-effective strategy is to defer further fetal echocardiograms after an initially inconclusive study, instead obtaining a definitive postnatal echocardiogram on all infants with concern remaining.
Prenatal diagnosis of congenital heart disease serves a critical role in parent education, prognostication, monitoring progression, and complication, as well as preparation for postnatal care. Coarctation of the aorta (CoA), which has a prevalence of four per 10,000 babies and represents 8% of congenital heart defects, is therefore a common perinatal concern among pediatric cardiologists. CoA is still the most common ductal-dependent cardiac defect missed at routine physical screening of the neonate; approximately 60%–80% of neonates with CoA remain unrecognized, presenting in low cardiac output states after spontaneous closure of the ductus arteriosus with resultant cardiovascular shock and death. Timely diagnosis and postnatal intervention are of critical importance as prenatal diagnosis of CoA has been shown to reduce morbidity and improve survival. However, CoA remains a difficult lesion to definitively diagnose by fetal echocardiogram (F-Echo) due to patency of the arterial duct and reliance on subjective measures such as ventricular asymmetry or dominance of the right heart structures. Ongoing efforts are being made to identify sensitive and specific quantitative prenatal echocardiographic markers, with most markers illustrating a poor positive predictive value. Still, there is a high false-positive rate for the fetal diagnosis of CoA, resulting in a large proportion of fetuses with multiple F-Echos during a pregnancy, initiation of prostaglandin E 1 following delivery, and immediate transfer to a cardiac unit for acquisition of a neonatal echocardiogram (N-Echo) to evaluate for CoA to further direct management.
While the benefits of prenatal diagnosis have been illustrated, repeating F-Echo after an initially inconclusive study may add expense but not add diagnostic clarity. Instead, the images may remain inconclusive, and providers must defer definitive diagnosis to the immediate postnatal period with acquisition of an N-Echo immediately after delivery. Little data are available to guide clinicians’ decisions in their diagnostic care of fetuses with prenatal concern for CoA and the ardency with which they should repeat F-Echos in lieu of deferring diagnosis until the acquisition of definitive N-Echo. Two recent studies employ cost-effectiveness analyses in understanding the utility of F-Echo in the diagnosis of congenital heart disease. We aimed to report on our single-center experience with prenatal evaluation and postnatal presentation of fetuses at risk for CoA along with the costs associated with their care. We hypothesized that a cost-utility analysis (CUA) could then be employed to improve the prenatal evaluation and postnatal management of fetuses with increased CoA risk by optimizing clinical outcomes while minimizing cost of care.
Methods
Data and Assumptions
A retrospective cohort study was performed and included all subjects at our institution between 2007 and 2014 with prenatal concern for CoA and/or who required surgical or transcatheter intervention for CoA in the first 30 days of life. The criteria for the prenatal diagnosis of CoA are both quantitative and qualitative. In our laboratory, we utilize aortic arch Z scores, flow acceleration through the arch, Doppler reversal of flow into the transverse arch, reversed foramen shunting, and novel measures as have been delineated in our laboratory. In addition, we look for subtle determinants of potential arch pathology including “splaying” of the distance between the left carotid and subclavian arteries, anatomic “tortuosity” of the aortic arch, and minor left heart pathology. In order to focus on a diagnosis of uncomplicated CoA, exclusion criteria included neonates with interventions for pathology unrelated to congenital heart disease during the hospitalization. Fetal subjects with no postnatal follow-up available were also excluded. Subjects in whom a prenatal diagnosis was made but whose postnatal evaluation revealed normal anatomy were only evaluated to quantify the cost of any postnatal hospitalization while verifying reassuring cardiac anatomy.
Subject records were reviewed to extract clinical variables. The term “extremis” was defined as initiation of inotropic support preoperatively in the first 24 hours of the hospitalization. Other cardiac diagnoses requiring intervention were defined as cardiac lesions that were intervened upon, either surgically or by catheter procedure during the first hospitalization. Aggregated costs were extracted from each individual’s medical record for the inpatient care beginning at their initial presentation and ending with subsequent discharge from that initial hospitalization. To capture societal expense, costs were defined as insurer reimbursement payments. Embedded in these hospitalization costs were all costs accrued during the initial hospitalization which would include preoperative intensive care unit (ICU) care, diagnostic imaging, surgical care, interventional catheter procedures, and the postoperative stay. The variable Standard Hospitalization represents the length of stay (LOS) and costs associated with those patients diagnosed prenatally or presenting clinically well. Extremis Hospitalization captures the experience and costs of those presenting in extremis. F-Echos with “high concern” were defined as those in which an N-Echo immediately after birth was advised. These data were aggregated and averaged to derive baseline assumptions for use in the CUA. Comparative statistics utilized a one-tailed Student’s t test. Seattle Children’s Hospital Institutional Review Board approved the study protocol.
To model prenatal costs that would not be captured in the hospitalization data, F-Echo costs were estimated using the sum of Common Procedural Terminology (CPT) codes 76825 (“Fetal echocardiogram 2D, complete”) and 76827 (“Doppler Echocardiogram, Fetal”). Similarly, each N-Echo cost was allocated costs associated with CPT code 93320 (“Echocardiogram 2D, Complete”) and CPT code 93325 (“Doppler Echocardiogram, Complete”). Each F-Echo appointment was also allocated costs associated with CPT code 99245 (“Outpatient Consult Clinic Visit, Level 5”). Costs were reported in 2014 U.S. dollars, with a 3% discount rate to standardize values and were a mix of private and public insurer.
Decision Analysis
The primary data from our institution were used to calculate cost and probability assumptions that were used in the decision tree analysis. Tree Age Pro 2014 (Williamstown, MA) was used in construction of the decision tool and performance of subsequent CUA. A decision tree was constructed to model four potential diagnostic approaches in a hypothetical patient population for whom initial obstetrical screening results in referral to a fetal cardiologist, and upon initial F-Echo the diagnosis of CoA remains equivocal ( Figure 1 ). The first paradigm, prenatal-multiple, modeled the approach predominantly followed by this institution in which patients whose F-Echo was initially inconclusive were subject to two additional F-Echo studies later in gestation and those with “highly concerning” F-Echo studies were evaluated by N-Echo immediately after birth. The second paradigm, postnatal-selective, modeled the pursuit of no further F-Echos but the obtaining of N-Echos on those neonates with a “highly concerning” initial F-Echo. Postnatal-all was the third paradigm and modeled the decision to pursue no further F-Echos after an initially inconclusive study but to obtain N-Echos on all patients regardless of the magnitude of concern on the initial F-Echo. The final paradigm that was included to establish a cost baseline, postnatal-none, assumed no further F-Echos nor N-Echos on any patients with initial equivocal F-Echo, instead relying on the postnatal physical exam to identify afflicted infants.
Expected value (EV) was calculated as the probability-weighted average cost for a paradigm. Quality-adjusted life years (QALYs) was the measure of effectiveness for each paradigm. In accordance with the Centers for Disease Control and Prevention, 78.8 years was the assumed life expectancy. The standard metric for CUA, the incremental cost-effectiveness ratio (ICER), was used to compare paradigms with a willingness-to-pay (WTP) threshold set at $50,000/QALY.
To test the durability of the conclusion, several one-way sensitivity analyses across reasonable clinical ranges were performed. In addition, a 1,000-patient Monte Carlo simulation was performed to subject the model to the influence of randomness. Assuming Gaussian distributed values between 0 and 1 at each chance node, this hypothetical cohort ran through the model and outcome frequencies were recorded.
Results
There were 115 subjects in whom there was prenatal or postnatal concern for CoA ( Figure 2 ). Of these, 18 did not have CoA postnatally, while 97 required intervention for CoA during the study period. Five patients who had undergone surgical interventions during the hospitalization for pathology unrelated to congenital heart disease were excluded. This left a final study population was 92 subjects. Of these, 31 were prenatally diagnosed with a “concern for CoA” through F-Echo. At our institution, multiple F-Echos after equivocal first F-Echo lead to a definitive diagnosis of CoA in 40% of fetuses. Of the 61 patients who were not diagnosed prenatally (mean age at diagnosis, 14.7 days), six were initially identified on asymptomatic postnatal screening (mean age at diagnosis, 4.8 days), and 35 were diagnosed with physical exam findings (mean age at diagnosis, 8 days). The remaining 20 subjects presented for care in extremis (mean age at diagnosis, 29.4 days). Of the 18 patients in whom there was a prenatal concern for CoA but who were confirmed postnatally to be without disease, nine were subject to a prolonged hospitalization while awaiting confirmation of normal cardiac anatomy (i.e., restriction of the ductus arteriosus). Six of these nine subjects were admitted to our hospital, allowing for analysis of clinical outcomes, cost, and LOS.
Population characteristics are outlined in Table 1 . The majority of subjects had CoA alone, with a minority having aortic arch hypoplasia or interrupted aortic arch. In those who were not diagnosed prenatally, extremis was the most common presentation (33%). Of those who presented postnatally but clinically well, murmur was the most common symptom (30%), followed by tachypnea (20%). There were 32 subjects who required additional cardiac interventions at the time of aortic arch repair, predominantly atrial or ventricular septal defect closure or patent ductus arteriosus ligation. Those patients presenting in extremis had longer ICU LOS (14.0 vs 9.3 days), total LOS (21.1 vs 16.8 days), and cost ($224,365 vs $187,494) incurred than those presenting in clinical stability. All-cause mortality was 1% (one of 92 in the study population), yet this was due to medical complexities other than would be directly attributable to the cardiac lesion.
| Demographics | N = 92 |
|---|---|
| Age at admission (days) | 11 |
| Presentation | |
| Prenatally diagnosed | 31 |
| Extremis | 20 |
| Murmur | 18 |
| Tachypnea | 12 |
| Diminished pulses | 5 |
| Screening given other abnormality | 4 |
| Failed pulse oximetry screen | 2 |
| Arch pathology | |
| CoA | 44 |
| Aortic arch hypoplasia | 25 |
| Both CoA and hypoplasia | 15 |
| Interrupted aortic arch | 8 |
| Other cardiac diagnosis requiring intervention | 32 |
| Interventions | |
| Surgical | 85 |
| Catheter (therapeutic) | 10 |
| Catheter (diagnostic) | 3 |
| Died | 1 |
Neonates with clinical concern for CoA who were admitted to the ICU for observation but were eventually shown to have normal cardiac anatomy had an average ICU LOS of 3.3 days with a cost of $29,780 during that time. None of these subjects had any additional cardiac pathology diagnosed postnatally.
Outcome data as well as average costs were used as baseline characteristics in the decision analysis for performance of the CUA ( Table 2 ). CUA revealed that postnatal-none was the least expensive paradigm, with an EV of $111,558, thus forming the cost baseline paradigm. However, this model also demonstrated the lowest effectiveness, with an incremental QALY 0.728 below postnatal-all. Postnatal-all was the superior model, with an effectiveness of 78.798 QALY and EV of $111,926, resulting in an ICER of $506/QALY above the baseline paradigm, significantly below the willingness-to-pay threshold. Prenatal-multiple, the approach currently followed by this institution, was the second most effective, with an incremental effectiveness 0.025 QALY less than postnatal-all, but was the most costly, with an EV of $156,200. These results are summarized in Table 3 . By convention, an ICER was not calculated for dominated paradigms, defined as those inferior to postnatal-all in terms both of cost and effectiveness.
| Decision analysis model inputs | Baseline | Citation | One-way sensitivity analysis range | One-way sensitivity analysis superior strategy |
|---|---|---|---|---|
| Probability | ||||
| N-Echo requested in “high concern” equivocal serial F-Echos | 0.93 | Primary data | ||
| CoA in “high concern” yet equivocal serial F-Echos | 0.68 | Primary data | ||
| CoA diagnosed clinically prior to onset of extremis | 0.67 | Primary data | 0.1–0.9 | No change |
| CoA in patient with inconclusive initial F-Echo (all patients) | 0.56 | Primary data | 0.1–0.9 | No change |
| Postnatal observation hospitalization required | 0.50 | Primary data | 0.1–0.9 | No change |
| CoA in “low concern” yet equivocal serial F-Echos | 0.45 | Primary data | ||
| Repeating F-Echo will lead to definitive prenatal diagnosis | 0.40 | Primary data | 0.1–0.9 | No change |
| Dying after presenting with onset of extremis | 0.05 | Primary data | 0.0–0.5 | Postnatal-none at < 0.001 |
| Cost | ||||
| Extremis hospitalization (including surgery) | $224,000 | Primary data | $50K–500K | No change |
| Standard hospitalization (including surgery) | $187,000 | Primary data | $50K–500K | No change |
| Observation hospitalization | $29,780 | Primary data | $2K–100K | No change |
| F-Echo complete + Doppler | $1,629 | Primary data | $200–2,000 | No change |
| N-Echo complete + Doppler | $654 | Primary data | $200-1,000 | No change |
| Fetal consult clinic visit | $199 | Primary data | $100–1,000 | No change |
| Utility | ||||
| Life without coarctation | 1.0 | |||
| Hospitalization in intensive care | 0.87 | Carroll et al. | 0.5–0.9 | No change |
| Death | 0.0 | |||
| Time | ||||
| Life expectancy (y) | 78.8 | Centers for Disease Control and Prevention | ||
| Length of extremis hospitalization (days) | 14.0 | Primary data | 5–30 | No change |
| Length of standard hospitalization (days) | 9.3 | Primary data | 5–30 | No change |
| Length of observation hospitalization (days) | 3.3 | Primary data | 1–7 | No change |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
