Effects of Milrinone on Neonates after Patent Ductus Arteriosus Ligation: A Retrospective Nationwide Database Study

Patent ductus arteriosus (PDA) – a major cardiovascular complication among preterm infants – is associated with mortality and severe comorbidities, including bronchopulmonary dysplasia (BPD), necrotizing enterocolitis (NEC), and retinopathy of prematurity (ROP) [1‒4]. Pharmacotherapy or surgery for PDA is required in 20–70% of very preterm infants [4‒7]; if pharmacotherapy with ibuprofen, indomethacin, or paracetamol is ineffective, PDA ligation is performed [1, 6, 8]. From 2006 to 2015, 2.9–8.4% of preterm infants in the USA required PDA ligation [7].

Postligation cardiac syndrome (PLCS) is a phenomenon of hemodynamic instability that manifests as postoperative hypoxemia and hypotension and requires prolonged respiratory support [9]. PLCS occurs in 7.2–44% of patients who undergo PDA ligation [10‒14] and can result in mortality, morbidities (such as BPD and NEC), and the need for home oxygen therapy [10, 13, 14]. Milrinone, a selective phosphodiesterase III inhibitor, is administered following PDA ligation for the prevention and treatment of PLCS [15‒18]. A retrospective study involving 52 patients who underwent PDA ligation reported that milrinone reduced the development of PLCS, catecholamine use, respiratory support, and steroid use [17]. Although some observational studies have reported that milrinone therapy after PDA ligation can reduce oxygenation or ventilation failure, cardiac function, and pharmacokinetics, there is little evidence regarding postoperative mortality, BPD, NEC, and ROP [18, 19]. Therefore, using a nationwide inpatient database in Japan, we aimed to explore the effectiveness of milrinone on in-hospital outcomes in infants who underwent PDA ligation.

This nationwide, retrospective cohort study was conducted using the Diagnostic Procedure Combination database. This database includes the administrative claims data and discharge abstracts of approximately 8,000,000 inpatients per year from >1,200 hospitals throughout Japan [20]. All 82 university hospitals in Japan are required to participate in the database, whereas the participation of community hospitals was voluntary. The database covers approximately 90% of hospitals with neonatal intensive care units (NICUs) in Japan [21]. The requirement for informed consent was waived due to the anonymity of the patient database, and the Institutional Review Board of the University of Tokyo approved this study (approval number: 3501-[5] [May 19th, 2021]).

The database comprised the following patient data: sex, gestational age, birth weight, main diagnoses, comorbidities on admission, and complications after admission recorded in Japanese text data and using the International Classification of Disease, Tenth Revision (ICD-10) codes indexed by the original Japanese procedure codes, patient transportation from other hospitals or in-hospital births, length of stay, discharge status, and hospitalization costs. Hospitalization costs were based on reference prices in the Japanese national fee schedule for item-by-item prices for surgical, pharmaceutical, laboratory, and other inpatient services. All abstract discharge data for each patient were recorded at discharge by attending physicians. Favorable sensitivities and specificities for diagnosis and procedure records in the database were demonstrated in several previous validation studies [22‒24]; additionally, several perinatal and neonatal studies were conducted using this database [25‒27].

We retrospectively identified patients admitted to the NICU who underwent PDA ligation between July 2010 and March 2021. The original Japanese procedure codes for PDA ligation and NICU admission were used to identify the patients. Patients who simultaneously underwent heart surgery other than PDA ligation were excluded. We divided patients into two groups to investigate the effectiveness of early-phase initiation of milrinone after PDA ligation. The milrinone group comprised patients who were administered milrinone on the first or second day after PDA ligation, and the non-milrinone group comprised those who were not administered milrinone after PDA ligation.

The primary outcomes were in-hospital morbidity (BPD, intraventricular hemorrhage [IVH], NEC, and ROP) and mortality. BPD was defined as oxygen use, noninvasive positive pressure ventilation, or ventilation at 36 weeks of corrected gestational age in preterm infants born under 32 weeks of gestational age. NEC was defined as undergoing abdominal surgery coded with the original Japanese procedure codes or diagnoses recorded with ICD-10 codes. ROP was defined as laser photocoagulation or antivascular endothelial growth factor treatment. Definitions of the ICD-10 codes for IVH and NEC are listed in Supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000533598). Secondary outcomes were postoperative steroid and blood transfusion use, discharge destination, post-discharge device use (including home oxygen therapy, tracheostomy, and home mechanical ventilation), length of hospital stay, and total hospitalization cost. We defined the currency exchange rate as 130 Japanese yen per 1 US dollar.

We examined the following patient characteristics: sex, gestational age, birth weight, in-hospital birth, general comorbidities, and the presence of neonatal asphyxia. Gestational age was categorized into four groups: <28 weeks (extremely preterm), 28–31 weeks (moderately preterm), 32–36 weeks (late preterm), and ≥37 weeks (term). Birth weight was categorized into four groups: <1,000 g (extremely low), 1,000–1,499 g (very low), 1,500–2,499 g (low), and ≥2,500 g (normal). General comorbidities were defined according to the pediatric complex chronic conditions (CCCs) classification system, version 2 [28], and categorized into four groups according to the number of CCCs (0, 1, 2, and ≥3). Additionally, we investigated the following treatment characteristics: preoperative drug use (cyclooxygenase inhibitors [indomethacin and ibuprofen], catecholamines [dopamine and dobutamine], adrenalin, milrinone, steroids, antibiotics, sedative agents [fentanyl, midazolam, phenobarbital, and dexmedetomidine], and muscle relaxants [rocuronium and vecuronium]), preoperative transfusions (red blood cells, platelets, fresh frozen plasma, albumin, and gamma-globulin), preoperative procedures (central venous catheterization, inhaled nitric oxide, intratracheal surfactant administration, intubation, and jaundice phototherapy), hospital type, hospital volume, and fiscal year. Regarding the hospital type, we investigated whether patients were admitted to a general perinatal medical center or a teaching hospital. The NICU in a general perinatal medical center in Japan is similar to level 3 NICUs in the USA [29]. The hospital volume was defined as the number of PDA ligations performed annually at each hospital and was categorized into tertiles (low, medium, and high). The fiscal years were categorized into two halves: 2010–2015 and 2016–2021. These background characteristics were adjusted in the subsequent analyses.

We conducted propensity score matching at a 1:4 ratio to compare outcomes between the two groups [30, 31]. We calculated the propensity scores of the patients and their treatment characteristics using a logistic regression model. Each patient in the milrinone group was matched with 4 patients with the closest estimated propensity scores within a caliper (≤0.2 of the pooled standard deviation) in the non-milrinone group. Nearest-neighbor matching with replacement was used to obtain matched samples by matching the subjects. Additionally, we calculated the standardized differences to examine the balance in the baseline characteristics of patients between the two groups for all patients, as well as for the 1:4 propensity score-matched paired cohort. A standardized difference of 10% implied a negligible difference between the two groups [31]. In the 1:4 propensity score-matched paired cohort, we performed a χ² test to compare proportions for categorical outcomes and the Kruskal-Wallis test to compare medians for continuous outcomes. A subgroup analysis of patients with a gestational age <28 weeks was also conducted because preterm infants with <28 weeks of gestational age often require PDA ligation and intensive cardiotropic support after PDA ligation [1, 2, 12].

For sensitivity analysis, we conducted an overlap propensity score-weighted analysis for the main and subgroup analyses to examine the robustness of our analysis. Overlap propensity score-weighted analysis is an extension of propensity score analysis that balances covariates between two groups and imitates a randomized trial without excluding most observations from the study sample [32, 33]. This analysis can compare highly comparable populations without being overly influenced by outliers that are almost always or never treated [33].

The threshold for significance was set at a level of 0.05, and all reported p values were two-sided. All statistical analyses were conducted using Stata/MP 17.0 (StataCorp, College Station, TX, USA).

We identified 4,117 patients who underwent PDA ligation in the NICU between July 2010 and March 2021. We excluded 297 patients who underwent other cardiac surgeries during PDA ligation. Of the 3,820 eligible patients, the milrinone and non-milrinone groups comprised 428 and 3,392 patients, respectively (Fig. 1).

Table 1 describes the background characteristics of all patients, as well as the 1:4 propensity score-matched patients. Before matching, the milrinone group was more likely to have a higher gestational age, birth weight, and number of CCCs than the non-milrinone group. After 1:4 propensity score matching, the milrinone and non-milrinone groups comprised 1,698 patients; the background characteristics were well-balanced between the two groups.

The crude outcomes of all patients are shown in Table 2. Among all patients, in-hospital morbidity occurred in 2,215 (58%) patients; 206 (5.4%) patients died. After 1:4 propensity score matching, no significant differences were observed in mortality (7.1 vs. 5.7%), in-hospital morbidity (55 vs. 50%), length of hospital stay (median: 113 vs. 102 days), hospitalization costs (median: approximately 86,000 vs. 82,000 US dollars), or other outcomes between the two groups. The results of the overlap propensity score-weighted analysis, wherein the weighted background characteristics were balanced (online suppl. Table 2), were comparable to those of the main analyses (Table 3).

In the subgroup analysis of patients born at <28 weeks of gestational age, the milrinone and non-milrinone groups comprised 224 and 2,209 patients, respectively (Fig. 1). After 1:4 propensity score matching (n = 221 vs. n = 882; online suppl. Table 3), the outcomes did not differ significantly, except for postoperative steroid use; the milrinone group received steroids after ligation more often than the non-milrinone group (68 vs. 60%; p = 0.025) (Table 4). In the overlapping propensity score-weighted analysis, no significant differences were observed in any outcomes, including steroid use (Table 3).

In this nationwide database study, we explored the effects of milrinone administration on short-term outcomes following PDA ligation using two propensity score analyses. Postoperative milrinone use was not significantly associated with reduced in-hospital morbidity or mortality. The subgroup analysis of patients born before 28 weeks of gestation demonstrated results similar to those in the main analysis.

PDA ligation can cause hemodynamic instability (i.e., PLCS) associated with severe morbidity [12, 13, 34]. Previous observational studies demonstrated improved cardiac function in patients who received milrinone after PDA ligation [15, 17‒19] as milrinone can reduce biventricular afterload [35]; however, whether milrinone therapy after PDA ligation improves clinical outcomes (i.e., mortality and severe morbidity) remains unclear. To our knowledge, the present study is the largest retrospective cohort study to compare the outcomes of neonates treated with and without milrinone after PDA ligation.

In the present study, milrinone was used in 11% of patients who underwent PDA ligation, while in a previous study of 85 patients who underwent PDA ligation, 19% were administered milrinone [16]. The patients in our study had a lower gestational age and birth weight than those in the previous study, and patients with a high gestational age may have tended to receive milrinone after ligation. Indeed, the milrinone group had a higher gestational age and birth weight than the non-milrinone group.

Our findings revealed that milrinone use after PDA ligation was not associated with decreased morbidity, mortality, discharge destination, post-discharge device use, length of hospital stay, or hospitalization costs. These findings are inconsistent with those of previous retrospective studies that reported the effectiveness of PDA ligation for hemodynamic instability [15, 17‒19]. However, these previous studies were conducted in a single institute, comprised a small number of patients, and did not adjust for clinical background [15, 17‒19]. The present study involved approximately 4,000 patients and was adjusted for patient backgrounds using propensity score matching. Furthermore, although previous studies have investigated postoperative cardiac function, hemodynamic instability, and pharmacokinetics [15, 17‒19], these outcomes may not have direct clinical relevance. The results of the present study suggest that milrinone administration after PDA ligation may be ineffective in improving short-term clinical outcomes. Clinicians should be aware of the uncertain efficacy regarding milrinone initiation after ligation.

In the subgroup analysis of patients with a gestational age <28 weeks, the results were almost consistent with those of the main analysis, excluding postoperative steroid use. That is, although a previous study reported the decreased administration of postoperative steroids in patients treated with milrinone, the milrinone group received steroids after ligation more often than the non-milrinone group [11]. Since milrinone is associated with decreasing blood pressure due to excessive vasodilation [11], steroids would have been used to raise blood pressure, specifically in preterm infants [36]. Milrinone use after ligation in preterm infants under 28 weeks of gestation may induce excessive vasodilation and thus require steroid administration.

This study had several limitations. First, we did not acquire laboratory data or imaging findings (e.g., echocardiography) to evaluate the severity of PDA and its complications. However, we adjusted for the clinical background, such as preoperative drug use and procedures, to estimate disease severity. Second, we did not describe the actual drug dosages administered to the patients because we were unable to obtain data on the partial use of drug vials, even though the effect of milrinone can differ by dosage [37]. Third, we were unable to obtain previous maternal histories, such as antenatal corticosteroid therapy, maternal comorbidities, and mode of delivery. Still, we believe that adjustment for neonatal conditions after birth can be a proxy for maternal factors. Fourth, we did not set outcomes specific to surgical complications as primary and secondary outcomes in this study. However, we consider that comprehensive outcomes are more relevant for patients and clinicians than outcomes specific to surgery. Therefore, the results of this study will be useful for patients and clinicians. Finally, we were unable to adjust for all confounding factors because this was a retrospective study. However, studies using real-world data can be superior to randomized controlled trials in the representativeness of the population and feasibility of conducting research [38]. Regardless of these limitations, this study is informative in that it describes the relevant outcomes of milrinone use after ligation in a large cohort. Further randomized controlled trials are warranted from the results of this study.

We performed propensity score analysis using a nationwide cohort of patients who underwent PDA ligation to compare the short-term outcomes of milrinone use after PDA ligation. Milrinone use after PDA ligation was not significantly associated with decreased in-hospital mortality or morbidities such as BPD, IVH, NEC, and ROP. Milrinone administration after PDA ligation may, therefore, be ineffective at improving short-term outcomes.

This study was approved by the Institutional Review Board of the University of Tokyo (approval number: 3501-[5] [May 19th, 2021]). The need for informed consent was waived by the Institutional Review Board of the University of Tokyo.

Takaaki Konishi received grants from Pfizer Co. Ltd., Kanzawa Medical Research Foundation, and Japan Kampo Medicines Manufacturers Association outside of the submitted work. There are no other conflicts of interest to disclose.

This work was supported by grants from the Ministry of Health, Labor, and Welfare, Japan (23AA2003 and 22AA2003 to H.Y.).

Hiroki Kitaoka and Takaaki Konishi conceptualized and designed the study, designed the data collection instruments, collected data, carried out the initial analyses, and drafted and revised the manuscript. Yoshihiko Shitara, Atsushi Ito, and Kohei Kashima conceptualized and designed the study and critically revised the manuscript. Asahi Fujita designed the data collection instruments, collected the data, carried out the initial analyses, and reviewed the manuscript. Hiroki Matsui coordinated and supervised data collection and reviewed the manuscript for important intellectual content. Motohiro Kato, Naoto Takahashi, and Hideo Yasunaga conceptualized and designed the study and critically reviewed and revised the manuscript for important intellectual content. All authors approved the final manuscript as submitted and agreed to be accountable for all aspects of the work.