Assessing the clinical significance of echocardiograms in determining treatment of patent ductus arteriosus in neonates

1 Introduction

Patent ductus arteriosus (PDA) is the persistent patency of the physiological ductus arteriosus between the descending thoracic aorta and pulmonary artery following birth [1]. Due to its inverse relationship with gestational age and birth weight, PDA is the most common cardiovascular pathology observed in preterm neonates, with a prevalence as high as 65% in neonates born at less than 26 weeks of gestation who survive to discharge [2–4]. The presence of a PDA in the neonate can be regarded as a continuum from being physiological to becoming pathological, especially with significant left-to-right shunting [4]. Determining which PDA is or will become pathological is extremely challenging and controversial. Often, the term hemodynamically significant PDA (hsPDA) is used to denote those PDAs that cause significant hemodynamic perturbation by way of ductal steal from systemic blood flow. However, there are no universally agreed upon criteria for this designation. As such, there remains much controversy regarding the timing and modality of treatment for a PDA, as well as the appropriate diagnosis of a hsPDA [3, 5]. With full term neonates, spontaneous ductus arteriosus closure occurs by 24 hours in 50% of neonates, 48 hours in 90%, and 72 hours in nearly all neonates [1, 3]. This is much less common in preterm neonates due to comorbidities caused by pulmonary over-circulation, reduced systemic perfusion, the inability for the musculature to fibrose in an effective physiological contraction and low response to constrictive effects of oxygen [5]. In preterm infants, PDA is associated with morbidities including kidney injury, chronic lung disease, intraventricular hemorrhage (IVH), and necrotizing enterocolitis (NEC) [1]. Appropriate diagnosis of a PDA is done through either clinical finding of murmur on auscultation or palpation of pulses, echocardiography (ECHO) or a combination of both [5]. Although various studies have shown that ECHO is the gold standard for PDA diagnosis, the utility of such a measure is often debated [3, 5–8], largely related to the fact that criteria that determine a pathological PDA leading to poor outcomes remains elusive [5]. There is also contradicting evidence regarding the long-term outcomes of infants who receive early therapeutic intervention for their PDA [3]. Despite a substantial reduction in IVH and a significant increase in PDA closure by indomethacin, there is minimal to no evidence of improved neurosensory or pulmonary outcomes at 18–21 months’ follow-up and even up to school age [9]. McNamara et al. suggest that this may be due to the fact that the traditional definition of a PDA does not take into account the physiological variability or magnitude of clinical effects that is attributable to the left-to-right ductal shunt that is seen [5]. McNamara et al. suggest that a better approach would be to “combine clinical markers of illness severity with ECHO-derived markers of hemodynamic disturbance” [5]. Despite this recommendation, the extent to which this has been implemented into current clinical practice is unknown. To this end, this study examines the extent to which clinicians use the ECHO and clinical parameters to guide the diagnosis and treatment of a PDA. We hypothesized that in the absence of firm ECHO criteria to determine a pathological PDA, clinical parameters are utilized in conjunction with ECHO findings to guide treatment decisions.

2 Methods

This was a retrospective case-control study conducted on all preterm neonates <37 weeks’ gestational age admitted to McMaster Children’s Hospital’s neonatal intensive care unit (NICU), a tertiary care hospital, with an ECHO confirmed PDA from January 2009 to January 2013, after obtaining institutional research ethics board approval. Any patient with a diagnosis of concurrent cardiac malformation other than a PDA were excluded from the analysis following data collection for hsPDA to decrease confounding variables. Gestational age was determined from the best estimate derived either by a reliable last menstrual period date, early obstetric ultrasound or New Ballard score [8]. ECHO reports, treatment measures for PDA and relevant demographic data were collected. For each subject, only one (the first) ECHO report completed by the pediatric cardiology service was utilized. Patients with no PDA on the first ECHO were excluded, while the remaining included cohort of subjects was classified into the following two groups a) having a hsPDA (exposure group) or b) non-hsPDA (control group). A hsPDA was defined as a PDA size >1.5 mm, left atrium-to-aorta ratio > 1.6, dilation of the left atrium (z-score >2), and dilation of the left ventricle (LV) (z-score >2) [1, 7].

Treatment status for each subject was then ascertained. Subjects within both exposure and control groups were classified into those who received a medical or surgical intervention and those who did not. Medical intervention was defined as indomethacin or ibuprofen administration within 48 hours of ECHO date [8]. Surgical ligation of a PDA was completed when repeated indomethacin/ibuprofen administration had failed or due to a contraindication to administer NSAIDs.

Demographic variables were compared between the two aforementioned subgroups of the cohort, including birth weight, antenatal corticosteroids (AC), method of birth (i.e caesarean section vs. vaginal), the 5 minute Apgar score, and whether surfactant was administered at birth. In addition, the outcome of interest –treatment of PDA –was comparatively assessed between the two groups. Finally, in an a priori planned separate analysis, variables deemed to have a clinical influence on treatment decisions –gestational age, presence of clinical murmur and blood pH levels on the day of the ECHO [7, 8] –were compared between subjects who received treatment versus those not treated in each of the two subgroups as well as in the entire cohort. Subgroup analyses of treatment status was performed evaluating the following subjects: a) those who met the ECHO criteria of LV enlargement with a z-score >2.0; b) those with a clinical murmur. All aforementioned comparisons were conducted using univariate analyses. Continuous variables were expressed as means (SD) and compared using one sided z-score and categorical variables were expressed as percentages and compared using variance and one sided z-score or two tailed t-test, as appropriate, unless otherwise indicated. A p value <0.05 was considered statistically significant. Box plots of birth weight (Fig. 1) between the two study groups of hsPDA and non-hsPDA as well as 5-minute Apgar score (Fig. 2) and gestational age (Fig. 3) show that there was a statistically significant difference between our study populations.

OPEN IN VIEWER

Fig. 1

Box plot of birth weights between the study groups of hsPDA and non-hsPDA.

OPEN IN VIEWER

Fig. 2

Box plot of 5-minute APGAR scores between the study groups of hsPDA and non-hsPDA.

OPEN IN VIEWER

Fig. 3

Box plot of gestational age between the study groups of hsPDA and non-hsPDA.

In the study period, there were 499 admissions to the NICU reviewed, of whom 446 received at least one ECHO. In 53 patients, there was no ECHO completed, the remaining 446 patients had an ECHO confirmed PDA on the first ECHO and were included as subjects in the study cohort. Following this, for data analysis, those who had congenital heart disease and a hsPDA but were not treated were excluded (1 patients), as the congenital heart disease may have been a confounder in the treatment decision. There were 95 subjects in the exposure (hsPDA) group and 350 subjects in the control (no hsPDA) group. Table 1 delineates the baseline demographic characteristics, as well as treatment status of the two subgroups. All of the baseline characteristics were deemed statistically significant between the two groups as measured by one-sided z-score analysis (Figs. 1–3). Of 95 patients with hsPDA confirmed on ECHO, 16 (16.8%) did not receive any treatment within 48 hours of the ECHO, while of 350 subjects with non-hs PDA on the ECHO, 39 (11.1%) received treatment within 48 hours.

Clinical characteristics deemed to be important towards decision making for treatment of PDA were compared in those who were and were not treated in each of the subgroups, as well as the entire cohort. These results are shown in Table 2.

The first subgroup analysis of the total study population showed that 3.1% (14 patients) were reported to have LV enlargement on ECHO. Within this subgroup 12 subjects (80.0%) had a hsPDA, of whom 11 (91.7%) were treated, while none of subjects with LV enlargement and non-hsPDA received treatment. Overall, 78.6% of patients with LV enlargement on ECHO received treatment. In the second subgroup limited to those with a clinical murmur (185 patients), there was 24.9% (46 patients) who had an ECHO finding in keeping with a hsPDA. Out of the same cohort of 185 patients with murmurs auscultated, 30.8% (57 patients) were treated based on the clinical finding. When these variables were combined, out of the 57 patients who received treatment, 82.5% (47 patients) had both a clinically auscultated murmur along with an ECHO finding of hsPDA. We then proceeded to analyze these results based on the impact of birth weight and gestational age. We conducted a two-tailed T-test on of unequal variances for birth weight between the subgroup with murmurs auscultated and the subgroup with no murmur auscultated and found a T-test value corresponding to a p-value of 0.468. We conducted the same analysis for gestational age between the two subgroups and found a p-value of 0.396.

Within the population of patients who received a positive ECHO finding for a hsPDA, 16.8% received no treatment. This demonstrates that physicians are not solely reliant on the ECHO, but rather use additional measures in guiding their decision making. An additional sub-cohort of patients, those who had a negative ECHO but received treatment for a hsPDA, confirm this finding –see Table 1. Within these two sub-cohorts, one could hypothesize that physicians relied on clinical markers rather than ECHO derived markers to guide their decision making.

Of the 185 patients with audible murmurs on auscultation, 57 (30.8%) were eventually treated for hsPDA based on the clinical finding. From the same cohort of 185 patients with a murmur, 47 patients (25.4%) received treatment based on the ECHO finding of hsPDA. When the two variables–clinically auscultated murmur and ECHO finding of hsPDA–were combined, the proportion of patients treated was 82.5% (47 out of 57 patients). Statistical analysis identifies that the combination of both murmur and ECHO derived markers act as predictors of treatment in hsPDA. This therefore suggests that when both the clinical finding of a murmur was combined with ECHO findings, the proportion of treated patients was greater than when each finding occurred on its own. When we conducted the two-tailed T-tests of unequal variances for the subgroups of clinically auscultated murmur or no murmur based on birth weight and gestational age, we found p-values of 0.468 and 0.396 as mentioned previously. These values therefore would indicate strong evidence that the two subgroups were not statistically different when it came to either gestational age or birth weight. Therefore, in regards to treatment decisions based on clinically auscultated murmurs, there is strong statistical evidence that these decisions were based on the finding of a murmur and not impacted by birth weight or gestational age as they were not statistically different.

Additional findings in the study show that 282 patients (approximately 56.5% of the sample size) were born before 36 weeks’ gestation. For 95.4% of this cohort of 282 patients, who received a positive echo, also received further treatment through either indomethacin or surgical ligation. This strongly supports the idea that the combination of early gestational age and ECHO markers is a key factor in guiding physician decision making regarding PDA treatment. LV enlargement was the last variable examined. Of the 14 patients with LV enlargement 12 were found to have a hsPDA. Of the 12 with a hsPDA and LV enlargement, 11 patients received treatment, illustrating that LV enlargement is a significant ECHO marker when diagnosing PDA. Hence, the results of the study show that gestational age, audible murmur, and ECHO-derived markers are most predictive in guiding physician decision making concerning the treatment of hsPDA. When the 16 patients in the study who received surgical ligation were analyzed it was seen that 12 of them received surgical intervention following a failed medical treatment. Of the remaining four patients, two of them had received surgical ligation six and 12 months following their birth and so were excluded based on no longer being within the neonatal period. The study showed only two patients received surgical ligation without first having any medical intervention for the PDA. One of the two patients required surgical ligation due to them having necrotizing enterocolitis, a contraindication to indomethacin use. The second patient had been born at full term and so underwent surgical ligation appropriately without an initial trial of indomethacin.

Of importance, this study demonstrated that 89.4% of neonates that were born in or admitted to a tertiary level hospital received at least one ECHO. This finding supports the fact that using an ECHO to assess neonates has become a standard of practice. Through this study it is apparent that clinical findings alone are not enough basis to rule in or out a significant PDA in order to make a clinical decision for treatment. Yet relying on ECHO alone may also not be sufficient and therefore the use of both classical clinical methods and ECHO, in conjunction, should be used.

Limitations in study design could explain some of the findings in this study. First, the study was retrospective which poses an inherent limitation, including the fact that some patient charts lacked pieces of data that were of interest to this study. There was also inherent inter-observer variability when it came to the echocardiogram reporting. The patients’ ECHOs were not all reported on by the same cardiologist. Thus, the reported significance of the PDAs may have differed based on reporter’s opinion. Furthermore, newborns will typically display physiological right-sided heart enlargement. Therefore, similar to previous studies, the current study only analyzed the left sided heart dilation for both the atrium and ventricle. This leads to the question that if there were atrio-septal defects (ASD) present as well as a PDA, one could expect that the left side measures may be within normal limits, as was seen in the majority of the patients in this study. Another limitation of the study was that during the analysis following data collection patients were excluded if they had a congenital heart disease in the group of patients who had hsPDA and not treated for the PDA. This was done in this group as not treating these patients who had ECHO findings in keeping with hsPDA may have been based on another variable of congenital heart disease. It was deemed that such issue would not have had an effect in the subgroups of treated patients in either hsPDA or non-hsPDA and so those patients were not excluded in those analyses. Future studies may be warranted to assess if treating in these patients with concurrent congenital heart disease has an effect. Lastly a limitation of the study is that the study was not powered to assess mortality in these two groups of patients and may have played a role in a patient not receiving treatment, though this was rarely encountered.

For confidentiality purposes the patients’ date of birth was not included in data collection and analysis. This limited the study in assessing whether the length of time from date of birth to the ECHO findings played a role in treatment decisions. It can be speculated that if the PDA is found earlier in the life of a patient, that less intensive management would be the primary treatment while a patient with a significant PDA that was discovered later in their early days of life may have a higher chance of being treated.

Although the study set out to assess the presence or absence of an overactive precordium on examination as a variable, it was noted that this data was often not reported on examination. Of the 499 patients, only four (4) patients’ charts reported the presence of absence of an overactive precordium. This may be due to the fact that the patients were in the NICU or very ill and therefore this portion of the exam was often overlooked. It may also be difficult to appreciate and comment on an exam finding such as overactive precordium if the patient was on ventilatory support. Additionally, the findings of overactive precordium, mechanical ventilation and clinical situations such as sepsis are all clinically relevant information which our study had not looked at which plays a role in daily decision making regarding treatment of these patients. These factors may play a confounding role in treatment and would be important for future studies to analyze. Though this study lends support to the notion that a combination of clinical parameters and ECHO findings should be used in conjunction when making the decision of whether or not to treat, and that in clinical practice this may not be the case. The fact that only four patients in the study had documentation that overactive precordium was examined is indicative that clinical exam parameters are often not completed.

The data reflect in Table 1 is limited in that not all the patient charts contained the complete information of interest to this study. First, within the treated group one (1) patient’s administration status of antenatal corticosteroids (AC) was unknown. Within the untreated group, three (3) ACs were unknown. The 5-minute Apgar scores were not mentioned in the reports of one (1) patient within the treatment group and two (2) patients within the non-treatment group. Lastly, the gestational week was not recorded in 30 patients in the non-treatment group.

In conclusion, ECHO derived markers remain the strongest overall predictors of the hsPDA. However, there is evidence that both audible murmur on auscultation as well as gestational age are clinical markers which help to further affirm or contest a physician’s inclination to treat the PDA. In a few instances, this study shows that physicians may contradict the ECHO derived recommendations and rely entirely upon their clinical judgment when diagnosing and treating PDAs. Therefore, despite ECHO being the gold standard, clinical examination and acumen play a significant role in the decision to treat patients with PDAs and remains an important tool.