Association of Patent Ductus Arteriosus Ligation With Death or Neurodevelopmental Impairment Among Extremely Preterm Infants Copyright 2017 American Medical Association. All rights reserved. Association of Patent Ductus Arteriosus Ligation With Death or Neurodevelopmental Impairment Among Extremely Preterm Infants Dany E. Weisz, MD, MSc; Lucia Mirea, PhD; Erin Rosenberg, BSc; Maximus Jang, BSc, MD; Linh Ly, MD; Paige T. Church, MD; Edmond Kelly, MD; S. Joseph Kim, MD, PhD; Amish Jain, MD; Patrick J. McNamara, MD, MSc; Prakesh S. Shah, MD, MSc IMPORTANCE Observational studies have associated patent ductus arteriosus (PDA) ligation among preterm infants with adverse neonatal outcomes and neurodevelopmental impairment in early childhood, with a resultant secular trend away from surgical treatment. However, to our knowledge, studies have inadequately addressed sources of residual bias, including survival bias and major neonatal morbidities arising before exposure to ligation. OBJECTIVE Evaluate the association between PDA ligation vs medical management and neonatal and neurodevelopmental outcomes. DESIGN, SETTING, AND PARTICIPANTS This retrospective cohort study of preterm infants younger than 28 weeks gestational age born between January 1, 2006, and December 31, 2012, with clinical and echocardiography diagnoses of hemodynamically significant PDA was conducted at 3 tertiary neonatal intensive care units and affiliated follow-up programs. EXPOSURE Surgical ligation vs medical management. MAIN OUTCOMES AND MEASURES The primary outcome was a composite of death or neurodevelopmental impairment (NDI) at 18 to 24 months corrected age. Secondary outcomes included death before discharge, NDI, moderate-severe chronic lung disease, and severe retinopathy of prematurity. Multivariable logistic regression analysis was used to adjust for perinatal and postnatal confounders. RESULTS Of 754 infants with hemodynamically significant PDA (mean [standard deviation] gestational age 25.7 [1.2] weeks and birth weight 813 [183] grams), 184 (24%) underwent ligation. Infants who underwent ligation had a higher frequency of morbidities before PDA closure, including sepsis, necrotizing enterocolitis, and a dependence on mechanical ventilation. After adjusting for perinatal characteristics and preligation morbidities, there was no difference in the odds of death or NDI (adjusted odds ratio (aOR), 0.83; 95% CI, 0.52-1.32), NDI (aOR, 1.27; 95% CI, 0.78-2.06), chronic lung disease (aOR, 1.36; 95% CI, 0.78-2.39) or severe retinopathy of prematurity (aOR, 1.61; 95% CI, 0.85-3.06). Ligation was associated with lower odds of mortality (aOR, 0.09; 95% CI, 0.04-0.21). CONCLUSIONS AND RELEVANCE Patent ductus arteriosus ligation among preterm neonates younger than 28 weeks gestational age was not associated with the composite outcome of death or NDI, and there were no differences in chronic lung disease, retinopathy of prematurity, or NDI among survivors. Mortality was lower among infants who underwent ligation, though residual survival bias could not be excluded. Previously reported associations of ligation with increased morbidity may be because of bias from confounding by indication. JAMA Pediatr. 2017;171(5):443-449. doi:10.1001/jamapediatrics.2016.5143 Published online March 6, 2017. Editorial page 422 Supplemental content Author Affiliations: Author affiliations are listed at the end of this article. Corresponding Author: Dany E. Weisz, MD, MSc, Department of Newborn and Developmental Paediatrics, Sunnybrook Health Sciences Centre, 2075 Bayview Ave, Toronto, ON M4N 3M5, Canada (dany.weisz@sunnybrook.ca). Research JAMA Pediatrics | Original Investigation (Reprinted) 443 Copyright 2017 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ by a Carnegie Mellon University User on 04/05/2021 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2017.0117&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 mailto:dany.weisz@sunnybrook.ca Copyright 2017 American Medical Association. All rights reserved. O ver the past decade, retrospective studies have asso-ciated patent ductus arteriosus (PDA) ligation with in-creased neonatal and neurodevelopmental morbid- ity, including chronic lung disease (CLD), retinopathy of prematurity (ROP), cerebral palsy, and cognitive, hearing, and visual impairments.1-7 These results have been associated with a secular trend toward a reduction in surgical ligation for per- sistent symptomatic PDA.8,9 However, ligation has also been associated with lower mortality compared with medical management.10 While the collective effect of previous studies has raised concern about surgical ligation, significant methodological shortcomings exist. To our knowledge, no contemporary ran- domized clinical trial has examined the effectiveness of PDA ligation. Importantly, previous observational studies investi- gating the association of PDA ligation and outcomes have only adjusted for antenatal and perinatal covariates in multivari- able analyses. While this approach may be sufficient to bal- ance confounders for interventions occurring shortly after birth, ligation often occurs weeks after birth. During the post- natal preligation period, infants may acquire multiple mor- bidities of prematurity (eg, sepsis and a dependence on inva- sive mechanical ventilation) that influence the decision to treat infants with ligation and outcomes. Failing to adjust for such confounders results in a high risk of residual bias against in- fants who underwent ligation because of confounding by in- dication and increased preligation illness severity. On the other hand, using ligation as a rescue treatment after the failure or contraindication of medical therapy confers a risk of survival bias in favor of infants who underwent ligation which has not, to our knowledge, been addressed in previous studies. The objective of this study was to evaluate the associa- tion of PDA ligation with neonatal and neurodevelopmental outcomes after accounting for antenatal, perinatal, and post- natal confounders. Methods We conducted a retrospective cohort study of extremely pre- term infants born at 27 weeks plus 6 days gestational age or younger and treated at 3 tertiary neonatal units in Toronto, Canada (Mount Sinai Hospital, Sunnybrook Health Sciences Centre, and the Hospital for Sick Children) from January 1, 2006, to December 31, 2012. Infants were included if they had a clinically significant PDA and had undergone at least 1 echo- cardiogram demonstrating a hemodynamically significant PDA, defined as a ductal diameter of 1.5 mm or larger. Infants with major congenital anomalies were excluded. This study was ap- proved by the respective research ethics boards from all 3 neo- natal intensive care unit institutions and the University of To- ronto, who also approved a waiver of informed consent. Management of PDA Treatment for PDA was at the discretion of the attending neo- natologist and typically occurred for a clinically and echocar- diographically significant PDA. Medical treatment aimed to- ward fac ilitating ductal closure (with indomethacin or ibuprofen) was used as first-line therapy. Pharmacotherapeu- tic treatment could be repeated at the discretion of the attend- ing team. Infants with hemodynamically significant PDA (HSPDA) were considered for surgical ligation after the fail- ure of, or contraindication to, medical therapy. The decision to refer an infant for PDA ligation was made by the attending neonatologist at each site in collaboration with a neonatologist with expertise in echocardiography. Before li- gation, infants at the 2 perinatal centers were transported to The Hospital for Sick Children for surgical ligation. The sur- gery was performed via a left lateral thoracotomy, an intra- or extrapleural approach, and the PDA was closed using either a clip or ligature at the discretion of the attending surgeon. Post- operative intensive care was supported by targeted neonatal echocardiography including the targeted administration of in- travenous milrinone for infants with critically low cardiac out- put in the immediate postoperative period.11 Outcomes and Assessment Surviving infants underwent neurodevelopmental assess- ments at 18 to 24 months corrected age, which consisted of a clinical examination, visual and hearing assessment, and cog- nitive evaluation using the Bayley Scales of Infant Develop- ment, Third Edition (BSID III), and/or the Ages and Stages Ques- tionnaire (ASQ). Clinical examinations and standardized motor assessments identified the presence of cerebral palsy, which was classified according to the Gross Motor Functional Clas- sification System.12 Cognition and language abilities were as- sessed predominantly using the BSID III, with a small num- ber of infants assessed exclusively using the ASQ. The primary outcome was a composite of death or mod- erate-severe neurodevelopmental impairment (NDI), evalu- ated at 18 to 24 months corrected age. Moderate-severe NDI was defined as a composite of neuromotor, neurocognitive, and/or neurosensory impairment (eTable 1 in the Supple- ment). Secondary outcomes included death before dis- charge, moderate-severe NDI, CLD (defined as treatment with supplemental oxygen or positive pressure support at 36 weeks Key Points Question Is patent ductus arteriosus ligation associated with adverse neonatal outcomes and neurodevelopmental impairment among extremely preterm infants? Findings In this cohort study of 754 extremely preterm infants, when postnatal preligation morbidities were properly accounted for, there was no difference in the composite outcome of death or neurodevelopmental impairment among infants who underwent ligation compared with those who were medically treated. While mortality was lower among infants who underwent ligation, there was no difference in neurodevelopmental impairment, chronic lung disease, or severe retinopathy or prematurity. Meaning Patent ductus arteriosus ligation may reduce mortality and is not associated with neurodevelopmental impairment. Previously reported associations of ligation with increased morbidity may be because of bias from confounding by indication rather than a detrimental causal effect of ligation. Research Original Investigation Outcomes of Patent Ductus Arteriosus Ligation in Extremely Preterm Infants 444 JAMA Pediatrics May 2017 Volume 171, Number 5 (Reprinted) jamapediatrics.com Copyright 2017 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ by a Carnegie Mellon University User on 04/05/2021 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 http://www.jamapediatrics.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 Copyright 2017 American Medical Association. All rights reserved. corrected gestational age) and severe ROP (defined as treat- ment with laser surgery or intravitreal vascular endothelial growth factor inhibitor). Data Sources and Collection The hospital records of all eligible infants were reviewed to ab- stract antenatal, neonatal, and outcome data. The echocardi- ography reports of each infant were reviewed to identify the infants with an echocardiographically significant PDA, de- fined as at least 1 echocardiogram with PDA diameter 1.5 mm or larger. The date and hemodynamic significance of the PDA in all echocardiograms were recorded from study reports to define the onset and longitudinal course (duration and severity) of each infant’s exposure to a ductal shunt (eTable 2 in the Supple- ment). The dates of echocardiography and/or clinical PDA clo- sure were recorded. For infants who underwent ligation, the date of PDA closure was recorded as the date of surgical liga- tion. For medically-treated infants, echocardiography clo- sure was recorded as the earliest date of the echocardiogram demonstrating ductal closure without a subsequent reopen- ing. Clinical closure was determined as the earliest date of the disappearance of established clinical signs of PDA (murmur, active precordium, and bounding pulses) in an infant in whom these signs were definitively present. The PDA of an infant was considered no longer hemodynamically significant if it under- went clinic al or echo closure or was “small” or “small- moderate” in size on echocardiography. Data on postnatal morbidities was collected for each day of life for all infants from birth until death or discharge from the neonatal intensive care unit. Specifically, the date of on- set of all morbidities was abstracted to include them as time- dependent covariates and characterize the timing of these morbidities in relation to ductal closure. These included in- traventricular hemorrhage and periventricular echogenicity, necrotizing enterocolitis stage 2 or higher, systemic dexameth- asone for the prevention or treatment of CLD (defined as a mini- mum 5-day course), seizures, systemic hypotension treated with inotropes, culture positive sepsis (defined as clinical sep- sis accompanied by the growth of a pathogenic organism from a sterile site), culture negative sepsis (defined as clinical sep- sis without a positive culture that was treated with at least 5 days of systemic antimicrobials), administration of inhaled ni- tric oxide, pneumothorax, and spontaneous intestinal perfo- ration. Daily information on respiratory support (invasive posi- tive pressure, noninvasive positive pressure, low flow oxygen, or none) and the average mean airway pressure was ab- stracted. Data were abstracted by study investigators at each site who collected the complete data from each infant's neo- natal intensive care unit course before reviewing neurodevel- opmental evaluations. For analyses, postnatal factors were considered time- dependent and were aggregated to represent cumulative ill- ness severity that was estimated for each day of life when in- fants were considered “at risk” for ligation. For infants who underwent ligation, the “at risk” period consisted of the days before surgical ligation. Medically-treated infants were con- sidered “at risk” for ligation during the period that the PDA he- modynamic significance was at least ”moderate.” This period for medically-treated infants with a persistent PDA was trun- cated at 60 days of age, as this was the latest postnatal day of surgery among the group who underwent ligation. Statistical Analysis Preliminary estimates indicated approximately 200 ex- tremely preterm infants were treated with ligation, and more than 600 infants received medical treatment for a clinically and echocardiographically significant PDA. Assuming 10% of infants would be lost to follow-up, we estimated that 180 in- fants who underwent ligation would have had the primary out- come assessed. Presuming a 60% event rate among the group who underwent ligation,1 a 2-sample, 2-sided test of propor- tions with 80% power and 5% type I error, using 180 infants who underwent ligation and 540 medically treated infants, was estimated to detect an 11% absolute difference in the primary outcome. Study infants were categorized according to whether they underwent surgical ligation (with or without prior pharmaco- therapy) or were not treated with ligation (“medically treated,” composed of cyclooxygenase inhibitor and/or conservative therapy). The distribution of perinatal and postnatal charac- teristics between groups who underwent ligation vs medical treatment was compared using the χ2 test for categorical vari- ables and the t test or Wilcoxon Rank-Sum test for continu- ous variables. The time to PDA closure for infants who under- went ligation vs medically treated infants was compared using the Kaplan-Meier analysis. The associations between PDA ligation and adverse out- comes were estimated using logistic regression analyses. Ini- tially, unadjusted analyses estimated crude odds ratios and 95% confidence intervals (CI) (model 1). Multivariable logistic re- gression analysis was used to adjust for possible confound- ing. A multivariable model was constructed using only ante- natal and perinatal covariates (model 2) to prov ide a comparison with the results of previous studies. The final model (model 3) included postnatal covariates representing morbidities that occurred during the period an infant was at risk of ligation. Variable selection for the final model was de- termined by backward elimination. Subcohort Analyses To explore the effect of survival bias (in which medically- treated infants had to survive to be eligible for ligation), subcohort analyses were conducted including only infants who survived with a HSPDA beyond specific time periods (day of life 3, 7, 14, 19, and 28). The longest periods of exclu- sion (19 and 28 days) were selected as they represented the earliest date of death of an infant in the group who under- went ligation (day of life 20) and the time period of major early morbidities of prematurity such as sepsis (day of life 28). An additional analysis was performed to explore the effect of possible information bias because of using the ASQ rather than the BSID among a small minority of infants. Sta- tistic al analyses were performed using SAS version 9.4 (SAS). All statistical tests were 2-sided with significance evaluated at the 5% level. Outcomes of Patent Ductus Arteriosus Ligation in Extremely Preterm Infants Original Investigation Research jamapediatrics.com (Reprinted) JAMA Pediatrics May 2017 Volume 171, Number 5 445 Copyright 2017 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ by a Carnegie Mellon University User on 04/05/2021 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 http://www.jamapediatrics.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 Copyright 2017 American Medical Association. All rights reserved. Results During the study period, 1040 preterm infants born at 27+6 ges- tational weeks or younger had a clinical diagnosis of PDA. Of these, 286 infants were excluded from the study (Figure 1). A total of 754 preterm infants had a clinical and echocardiography di- agnosis (≥1.5mm) of PDA, of whom 570 received medical treat- ment only and 184 were treated with surgical ligation. The pri- mary outcome of death or NDI at 18 to 24 months corrected age was known for 680 infants (90.2%). Compared with the infants who survived to discharge with known neurodevelopmental out- comes (n= 542), the surviving infants who were lost to follow- up (n = 74) were more likely to have been singleton and less likely to have been treated at Sunnybrook Health Sciences Centre. How- ever, there were no differences in other perinatal characteristics, postnatal morbidities, or neonatal outcomes of CLD and severe ROP (eTable 3 in the Supplement). Patient Characteristics The antenatal and perinatal characteristics of medically and surgically treated infants are presented in Table 1. Infants treated with ligation had lower gestational age and birth weight and were more likely to be female, require intensive delivery room resuscitation, and receive indomethacin prophylaxis. In- fants who underwent ligation had a longer exposure to an HSPDA, with most medically-treated infants experiencing duc- tal closure prior to the timing of ligation in surgically treated infants (eFigure 1 in the Supplement). Infants who underwent ligation had higher rates of mor- bidity during the postnatal period of ductal patency com- pared with medically treated infants. (Table 2) The burden of invasive mechanical ventilation was higher among infants who underwent ligation, who required a significantly higher aver- age daily mean airway pressure (9.6 cm H2O [±1.9] vs 7.9 cm H2O [±2.3]; 95% CI, 3.8-4.6), mean difference (1.7 cm H2O; 95% CI, 1.3-2.1), and more days of invasive mechanical ventilation Table 1. Antenatal and Perinatal Characteristics of the Cohort of Extremely Preterm Infants With Clinically and Echocardiographically Significant Patent Ductus Arteriosus Characteristic No. (%) Difference (95% CI) P Valuea Surgical Ligation (n = 184) Medical Treatment Only (n = 570) GA, mean (SD), wk 25.2 (1.1) 25.9 (1.2) −0.7 (−0.87 to −0.48) <.001 BW, mean (SD), g 742 (163) 835 (183) −93 (−125 to −66) <.001 SGA, <10th percentile 29 (15.8) 63 (11.0) 4.8% (−1.1 to 10.6) .09 Initial level 3 hospital NA .39 SB 74 (25.5) 216 (74.5) MSH 79 (22.4) 273 (77.6) HSC 31 (27.7) 81 (72.3) Multiple gestation 54 (29.3) 200 (35.1) −6.8% (−13.4 to 1.9) .19 Male 86 (46.7) 322 (56.5) −9.8% (−18.0 to −1.4) .02 Antenatal corticosteroids, full 95 (51.6) 317 (55.6) −4.0% (−12.3 to 4.3) .38 Vaginal delivery 94 (51.1) 286 (50.2) 0.9% (−7.4 to 9.2) .87 Not born at institution 56 (30.4) 161 (28.2) 2.2% (−5.4 to 9.8) .58 Intensive delivery room resuscitationb 181 (98.4) 531 (93.2) 5.2% (2.4-8.0) .007 5-min Apgar <7 78 (42.4) 217 (38.1) 4.3% (−3.0 to 12.5) .30 Indomethacin prophylaxis 48 (26.1) 101 (17.7) 8.4% (1.3-15.4) .02 SNAPII score ≥20 73 (39.7) 187 (32.8) 6.9% (−1.2 to 14.9) .09 Abbreviations: BW, birth weight; GA, gestational age; HSC, Hospital for Sick Children; MSH, Mount Sinai Hospital; NA, not applicable; SB, Sunnybrook Health Sciences Centre; SGA, small for gestational age; SNAPII, score for neonatal acute physiology II. a t-test or Wilcoxon test for continuous data, or χ2 test for categorical data. b Defined as intubation, chest compressions, or epinephrine administration. Figure 1. Flow Diagram of Infants Included in the Study 1040 Infants (GA ≤ 27+6) with diagnosis of PDA 754 Clinical and echo diagnosis of significant PDA 570 Medical treatment only 286 Infants excluded 37 No echo performed 224 PDA < 1.5mm on all echo 11 Congenital anomaly 14 Incomplete data 184 Surgical ligation 80 Conservative management 490 NSAID treatment 77 Follow-up complete 431 Follow-up complete 166 NSAID treatment and ligation 18 Primary ligation 155 Follow-up complete 17 Follow-up complete GA indicates gestational age; NSA, nonsteroidal anti-inflammatory drug; and PDA, patent ductus arteriosus. Research Original Investigation Outcomes of Patent Ductus Arteriosus Ligation in Extremely Preterm Infants 446 JAMA Pediatrics May 2017 Volume 171, Number 5 (Reprinted) jamapediatrics.com Copyright 2017 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ by a Carnegie Mellon University User on 04/05/2021 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 http://www.jamapediatrics.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 Copyright 2017 American Medical Association. All rights reserved. (26.9 [ ± 11.0] vs 14.1 [± 13.1]; 95% CI, 22.0-28.2), mean differ- ence (12.7 days; 95% CI, 10.7-14.8) in the postnatal period be- fore ductal closure (Figure 2). Main Outcomes On univariable analysis (model 1), ligation was associated with lower mortality but increased CLD, ROP and NDI, and there was no difference in the composite outcome of death or NDI (Table 3). After adjusting for antenatal and perinatal confounders only (model 2), the associations between ligation and NDI, CLD and ROP were attenuated but remained significant (Table 3). Liga- tion remained associated with lower mortality and there was no difference in the composite outcome of death or NDI. However, after further adjustment for postnatal, preductal closure confounders (model 3), the associations between surgi- cal ligation and NDI (adjusted odds ratio [aOR] 1.27; 95% CI, 0.78- 2.06), CLD (aOR 1.36; 95% CI, 0.78-2.39), and ROP (aOR 1.61; 95% CI, 0.85-3.06) were no longer significant (Table 3). Ligation re- mained associated with lower mortality (aOR 0.09; 95% CI, 0.04- 0.21). and there was no significant association with the compos- ite outcome of death or NDI (aOR 0.83; 95% CI, 0.52-1.32). The causes and timing of death (n = 17 in the ligation group; n = 121 in the medically-treated group) are reported in eTable 4 in the Supplement. Among those who died, medically-treated infants had higher rates of sepsis and early respiratory failure, and mortality occurred earlier in the postnatal period. Subcohort Analyses In the subcohort analyses (to address survival bias) that in- cluded infants who survived beyond a specific age, ligation was associated with lower mortality among all subcohorts and re- duced odds of the composite outcome of death/NDI in the Figure 2. Average Daily Mean Airway Pressure Over the First 40 Days of Life 10.5 10.0 M ea n A ir w ay P re ss ur e, c m H 2O Postnatal Age, d 9.5 9.0 8.5 8.0 7.5 7.0 0 20 30 4010 15 25 355 MedianQ1 Date of ligation Q3 Ligation No ligation Average daily mean airway pressure (cm H2O) with 68% and 95% CIs over the first 40 days of life for medically-treated infants (black line) vs infants who underwent ligation (orange line) before ductal closure. Infants no longer contributed data after the date of ductal closure, leading to the widening of CIs over time as the number of infants with persistent hemodynamically significant patent ductus arteriosus diminished with time. The median date of ligation was day of life 29, with the interquartile range (day of life 22 to day of life 38) (solid gray box). The earliest date of ligation was on day of life 7. Table 2. Morbidity Arising During the NICU Course, Period of Ductal Patency (Before Surgical Ligation or Medical Closure), and At-Risk Period for Surgical Ligation Among the Infant Cohort (N = 754) Morbidity No. (%) P ValueaSurgical Ligation (n = 184) Medical Treatment (n = 570) Incidence of Morbidity Before Discharge (1) Incidence of Morbidity Before Ligation, (2) Incidence of Morbidity Before Discharge (3) Incidence of Morbidity Before Medical PDA Closure (4) Incidence of Morbidity Before Medical Closure or DOL 60 (5) (1) vs (3) (2) vs (4) (2) vs (5) Culture positive sepsis 131 (71.2) 91 (49.5) 270 (47.4) 172 (30.2) 168 (29.5) <.001 <.001 <.001 Culture negative sepsis 107 (58.1) 92 (50.0) 255 (44.7) 225 (39.5) 225 (39.5) .001 .01 .01 Grade 3/4 IVH 31 (16.8) 31 (16.8) 128 (22.5) 128 (22.5) 111 (19.5) .10 .10 .43 Hypotension inotropes 59 (32.1) 54 (29.3) 150 (26.3) 128 (22.4) 128 (22.4) .13 .06 .06 NEC, Stage ≥2 22 (12.0) 17 (9.2) 51 (8.9) 29 (5.1) 28 (4.9) .23 .04 .03 Pneumothoraxb 11 (6.0) 7 (3.8) 23 (4.0) 22 (3.9) 22 (3.9) .27 .97 .97 Nitric oxide 9 (4.9) 6 (3.3) 32 (5.6) 27 (4.7) 27 (4.7) .71 .39 .39 Seizure 23 (12.5) 17 (9.2) 51 (8.9) 41 (7.2) 41 (7.2) .14 .37 .37 SIP 9 (4.9) 9 (4.9) 10 (1.7) 9 (1.6) 9 (1.6) .02 .01 .01 Dexamethasonec 31 (16.8) 8 (4.3) 37 (6.5) 12 (2.1) 12 (2.1) <.001 .11 .11 Abbreviations: DOL, day of life; IVH, intraventricular hemorrhage; NEC, necrotizing enterocolitis; NICU, neonatal intensive care unit; PDA, patent ductus arteriosus; SIP, spontaneous intestinal perforation. a χ2 test or Fisher exact test. b Pneumothorax requiring thoracostomy tube insertion. c Systemic dexamethasone administered for the prevention or treatment of chronic lung disease, minimum 5 d course. Outcomes of Patent Ductus Arteriosus Ligation in Extremely Preterm Infants Original Investigation Research jamapediatrics.com (Reprinted) JAMA Pediatrics May 2017 Volume 171, Number 5 447 Copyright 2017 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ by a Carnegie Mellon University User on 04/05/2021 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 http://www.jamapediatrics.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 Copyright 2017 American Medical Association. All rights reserved. subcohorts of infants with persistent HSPDA beyond the 20th day of life (eTable 5 in the Supplement). The association of PDA ligation and death/NDI (aOR 0.86; 95% CI, 0.51-1.44) or NDI alone (aOR 1.42; 95% CI, 0.83-2.43) in the subcohort of infants evaluated at neurodevelopmental follow-up using the BSID only (n = 665, 158 underwent liga- tion) was similar to results from the primary analysis (which included infants evaluated using the BSID or ASQ). Discussion In this retrospective cohort study of extremely preterm infants with HSPDA, surgical ligation was not associated with higher odds of the composite outcome of death or NDI, compared with medi- cal management alone. Ligation was also not associated with NDI, CLD, or severe ROP; however, PDA surgery was associated with reduced odds of mortality. These findings contrast with multiple large cohort studies over the past decade that have strongly as- sociated surgical ligation with NDI, CLD, and ROP.1-4,13,14 Methodological differences likely account for the diver- gence in results. Previous studies did not adjust for postna- tal, preligation confounders, which represent increased ill- ness severity. We observed similar findings as previous studies when we only included antenatal and perinatal covariates (model 2) which changed with the inclusion of postnatal co- variates measured before ductal closure (model 3). The at- tenuation of the associations with the inclusion of the post- natal covariates suggests that bias because of confounding by indication (as sicker infants are more likely to be referred for ligation), rather than a detrimental causal effect of the sur- gery itself, may explain the previously reported associations of ligation and adverse outcomes. Local clinical practices may also have contributed to the absence of an association between ligation and adverse out- comes. Targeted neonatal echocardiography was used to guide management aimed at ductal closure, including the use of a delayed, selective approach in referring infants for surgical li- gation, a clinical practice which has been associated with im- proved neonatal and neurodevelopmental outcomes.14,15 In this study, mortality was significantly lower for infants who underwent ligation compared with medically-treated in- fants, which corroborates previous studies.2,10 Although this finding is compelling, the association between ligation and lower mortality is, like prior studies, possibly influenced by survival bias or confounding by contraindication (in which in- fants with the highest illness severity are considered unsuit- able for surgery and do not survive). For example, medically- treated infants were more likely to have a cerebral injury (eTable 4 in the Supplement), from which several died early in life af- ter the withdrawal of life-sustaining medical therapy. In ad- dition, medically-treated infants were more likely to die from sepsis, suggesting that increased illness severity may have ren- dered them too unstable for PDA surgery. To reduce survival bias, we conducted subcohort analyses excluding infants who died or experienced PDA closure within the first weeks of life, though infants who underwent ligation continued to have significantly lower mortality compared with medically-treated infants. However, these findings should be interpreted with caution because of the possibility of residual survival bias. For example, it is unknown for an infant with a large PDA who died of sepsis on the28th day of life whether prior PDA ligation may have reduced the mortality risk. The poten- tial for residual survival bias means that the association of li- gation with outcomes that include mortality requires further evaluation before clinicians can be confident that ligation im- proves survival. Ultimately, the effect of survival bias and an es- timate of any mortality benefit of ligation may only be reliably evaluated in a randomized clinical trial. The strengths of this study include the in-depth abstrac- tion of postnatal morbidities and indices of illness severity, such as daily respiratory support, in a large cohort of extremely pre- term infants. This data collection included key covariates in the multivariable analyses to minimize confounding by indi- Table 3. Neonatal and Neurodevelopmental Outcomes of Infants Who Underwent Ligation vs Medically Treated Infantsa Outcome No. (%) Model 1: Crude OR (95% CI) Model 2: AOR (95% CI)b Antenatal/Perinatal Covariates Only Model 3: AOR (95% CI)c Model 2 and Postnatal, Preductal Closure Covariates Ligation (n = 184) Medical Treatment (n = 570) Death or moderate-severe neurodevelopmental impairmentd 110 (59.8) 299 (52.5) 1.24 (0.87-1.77) 0.97 (0.65-1.44) 0.83 (0.52-1.32) Death before dischargee 17 (9.2) 121 (21.2) 0.38 (0.22-0.65) 0.17 (0.09-0.31) 0.09 (0.04-0.21) Moderate-severe neurodevelopmental impairment 92 (50.0) 174 (30.5) 1.79 (1.22-2.62) 1.64 (1.08-2.51) 1.27 (0.78-2.06) Chronic lung disease 141 (76.6) 237 (41.6) 3.67 (2.44-5.52) 3.13 (1.96-5.00) 1.36 (0.78-2.39) Severe retinopathy of prematurity 42 (22.8) 30 (5.3) 4.48 (2.70-7.45) 2.67 (1.52-4.68) 1.61 (0.85-3.06) Abbreviations: AOR, adjusted odds ratio; GA, gestational age; NEC, necrotizing enterocolitis; NICU, neonatal intensive care unit; OR, odds ratio; SGA, small for gestational age; SNAP, score for neonatal acute physiology. a Reference is medically treated infants. b Adjusted for GA, SGA, antenatal corticosteroids, sex, multiple gestation, a SNAP 2 score of 20 or higher, and treatment center. c Adjusted for covariates in model 2 plus the following morbidities if they occurred before PDA closure: culture positive sepsis, severe intraventricular hemorrhage, inotrope use, NEC stage 2 or greater, average daily mean airway pressure, proportion of days of invasive mechanical ventilation, total dose of indomethacin, and systemic dexamethasone. d Primary composite outcome includes death before 18 to 24 mo neurodevelopmental follow-up. e Death before discharge from the NICU. Research Original Investigation Outcomes of Patent Ductus Arteriosus Ligation in Extremely Preterm Infants 448 JAMA Pediatrics May 2017 Volume 171, Number 5 (Reprinted) jamapediatrics.com Copyright 2017 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ by a Carnegie Mellon University User on 04/05/2021 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 http://www.jamapediatrics.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 Copyright 2017 American Medical Association. All rights reserved. cation. In addition, the comprehensive collection of all echo- cardiography data permitted identification of the “at-risk” pe- riod for ligation for each infant based on an objective estimate rather than clinician preference or perception. Finally, the pro- portion of infants lost to neurodevelopmental follow-up was low (9.8%)16 and these infants had similar characteristics as those for whom follow-up was complete, increasing the con- fidence in our estimates (eTable 3 in the Supplement). Limitations This study is limited by possible ascertainment bias, as medi- cally-treated infants were considered eligible for ligation until the date of confirmed PDA closure, which may have been de- layed relative to the true date of ductal closure. This may have led to overestimation of the “at-risk of ligation” period for some medically-treated infants and introduced bias in favor of the medically-treated group, as all morbidities that occurred be- fore this date were adjusted for in the multivariable analyses. In addition, a small minority of infants were evaluated using the ASQ rather than the BSID for the cognition and language as- sessments. However, results from a subcohort analysis includ- ing only infants evaluated using the BSID were similar to those from the full cohort, suggesting minimal ascertainment bias ow- ing to variable neurodevelopmental testing methods. Finally, there remains the potential for confounding by contraindica- tion, residual survival bias, and residual confounding because of unmeasured covariates in this analysis. Conclusions Compared with medical management, surgical ligation of HSPDA is associated with lower mortality without increased CLD, ROP, NDI, or the composite of death or NDI. The previously reported associations of ligation with adverse neonatal outcomes and neu- rodevelopmental impairment may be because of confounding by indication rather than a detrimental causal effect of PDA sur- gery. However, well-designed randomized clinical trials are needed to evaluate the relative effects of medical and surgical treatment of HSPDA on neonatal and early childhood outcomes. ARTICLE INFORMATION Accepted for Publication: December 9, 2016. Published Online: March 6, 2017. doi:10.1001/jamapediatrics.2016.5143 Author Affiliations: Department of Newborn and Developmental Pediatrics, Sunnybrook Health Sciences Centre, Toronto, Canada (Weisz, Church); Department of Pediatrics, University of Toronto, Toronto, Canada (Weisz, Ly, Church, Kelly, Jain, McNamara, Shah); Phoenix Children's Hospital, Phoenix, Arizona (Mirea); Department of Pediatrics, Hospital for Sick Children, Toronto, Canada (Rosenberg, Ly, McNamara); Department of Pediatrics, Mt. Sinai Hospital, Toronto, Canada (Jang, Kelly, Jain, Shah); Department of Medicine, University of Toronto, Toronto, Canada (Kim); Department of Physiology, University of Toronto and Physiology and Experimental Medicine, SickKids Research Institute, Toronto, Canada (McNamara); Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Canada (Shah). Author Contributions: Dr Weisz had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: Weisz, Mirea, Ly, Kim, McNamara, Jain, Shah, Church. Acquisition, analysis, or interpretation of data: All authors. Drafting of the manuscript: Weisz, Mirea, Shah. Critical revision of the manuscript for important intellectual content: All authors. Statistical analysis: Weisz, Mirea. Administrative, technical, or material support: Weisz, Shah. Supervision: Mirea, Ly, Kim, Jain, McNamara, Shah. Conflict of Interest Disclosures: None reported. Additional Contributions: The authors thank Dr William Benitz, MD, Stanford University, for his review of the study results and insightful comments. Dr Benitz was not compensated for his contributions. REFERENCES 1. Kabra NS, Schmidt B, Roberts RS, Doyle LW, Papile L, Fanaroff A; Trial of Indomethacin Prophylaxis in Preterms Investigators. Neurosensory impairment after surgical closure of patent ductus arteriosus in extremely low birth weight infants: results from the Trial of Indomethacin Prophylaxis in Preterms. J Pediatr. 2007;150(3):229-234, 234.e1. 2. Madan JC, Kendrick D, Hagadorn JI, Frantz ID III; National Institute of Child Health and Human Development Neonatal Research Network. Patent ductus arteriosus therapy: impact on neonatal and 18-month outcome. Pediatrics. 2009;123(2):674-681. 3. Mirea L, Sankaran K, Seshia M, et al; Canadian Neonatal Network. Treatment of patent ductus arteriosus and neonatal mortality/morbidities: adjustment for treatment selection bias. J Pediatr. 2012;161(4):689-94.e1. 4. Chorne N, Leonard C, Piecuch R, Clyman RI. Patent ductus arteriosus and its treatment as risk factors for neonatal and neurodevelopmental morbidity. Pediatrics. 2007;119(6):1165-1174. 5. Bourgoin L, Cipierre C, Hauet Q, et al. Neurodevelopmental outcome at 2 years of age according to patent ductus arteriosus management in very preterm infants. Neonatology. 2016;109(2): 139-146. 6. Janz-Robinson EM, Badawi N, Walker K, Bajuk B, Abdel-Latif ME; Neonatal Intensive Care Units Network. Neurodevelopmental outcomes of premature infants treated for patent ductus arteriosus: a population-based cohort study. J Pediatr. 2015;167(5):1025-1032, e3. 7. Clyman R, Cassady G, Kirklin JK, Collins M, Philips JB III. The role of patent ductus arteriosus ligation in bronchopulmonary dysplasia: reexamining a randomized controlled trial. J Pediatr. 2009;154(6):873-876. 8. Lokku A, Mirea L, Lee SK, Shah PS; Canadian Neonatal Network. Trends and outcomes of patent ductus arteriosus treatment in very preterm infants in Canada [published online September 20, 2016]. Am J Perinatol. 9. Hagadorn JI, Brownell EA, Trzaski JM, et al. Trends and variation in management and outcomes of very low-birth-weight infants with patent ductus arteriosus. Pediatr Res. 2016;80(6):785-792. 10. Weisz DE, More K, McNamara PJ, Shah PS. PDA ligation and health outcomes: a meta-analysis. Pediatrics. 2014;133(4):e1024-e1046. 11. Jain A, Sahni M, El-Khuffash A, Khadawardi E, Sehgal A, McNamara PJ. Use of targeted neonatal echocardiography to prevent postoperative cardiorespiratory instability after patent ductus arteriosus ligation. J Pediatr. 2012;160(4):584-589.e1. 12. Palisano R, Rosenbaum P, Walter S, Russell D, Wood E, Galuppi B. Development and reliability of a system to classify gross motor function in children with cerebral palsy. Dev Med Child Neurol. 1997; 39(4):214-223. 13. Clyman RI. Surgical ligation of the patent ductus arteriosus: treatment or morbidity? J Pediatr. 2012; 161(4):583-584. 14. Wickremasinghe AC, Rogers EE, Piecuch RE, et al. Neurodevelopmental outcomes following two different treatment approaches (early ligation and selective ligation) for patent ductus arteriosus. J Pediatr. 2012;161(6):1065-1072. 15. Jhaveri N, Moon-Grady A, Clyman RI. Early surgical ligation versus a conservative approach for management of patent ductus arteriosus that fails to close after indomethacin treatment. J Pediatr. 2010;157(3):381-387, 387.e1. 16. Kristman V, Manno M, Côté P. Loss to follow-up in cohort studies: how much is too much? Eur J Epidemiol. 2004;19(8):751-760. Outcomes of Patent Ductus Arteriosus Ligation in Extremely Preterm Infants Original Investigation Research jamapediatrics.com (Reprinted) JAMA Pediatrics May 2017 Volume 171, Number 5 449 Copyright 2017 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ by a Carnegie Mellon University User on 04/05/2021 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 http://jama.jamanetwork.com/article.aspx?doi=10.1001/jamapediatrics.2016.5143&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143 https://www.ncbi.nlm.nih.gov/pubmed/17307535 https://www.ncbi.nlm.nih.gov/pubmed/17307535 https://www.ncbi.nlm.nih.gov/pubmed/19171637 https://www.ncbi.nlm.nih.gov/pubmed/22703954 https://www.ncbi.nlm.nih.gov/pubmed/22703954 https://www.ncbi.nlm.nih.gov/pubmed/17545385 https://www.ncbi.nlm.nih.gov/pubmed/26726863 https://www.ncbi.nlm.nih.gov/pubmed/26726863 https://www.ncbi.nlm.nih.gov/pubmed/26227439 https://www.ncbi.nlm.nih.gov/pubmed/26227439 https://www.ncbi.nlm.nih.gov/pubmed/19324366 https://www.ncbi.nlm.nih.gov/pubmed/19324366 https://www.ncbi.nlm.nih.gov/pubmed/27649293 https://www.ncbi.nlm.nih.gov/pubmed/27649293 https://www.ncbi.nlm.nih.gov/pubmed/27649293 https://www.ncbi.nlm.nih.gov/pubmed/27649293 https://www.ncbi.nlm.nih.gov/pubmed/27509008 https://www.ncbi.nlm.nih.gov/pubmed/24639268 https://www.ncbi.nlm.nih.gov/pubmed/22050874 https://www.ncbi.nlm.nih.gov/pubmed/9183258 https://www.ncbi.nlm.nih.gov/pubmed/9183258 https://www.ncbi.nlm.nih.gov/pubmed/22795223 https://www.ncbi.nlm.nih.gov/pubmed/22795223 https://www.ncbi.nlm.nih.gov/pubmed/22795222 https://www.ncbi.nlm.nih.gov/pubmed/20434168 https://www.ncbi.nlm.nih.gov/pubmed/20434168 https://www.ncbi.nlm.nih.gov/pubmed/15469032 https://www.ncbi.nlm.nih.gov/pubmed/15469032 http://www.jamapediatrics.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jamapediatrics.2016.5143