530 - Data-Driven Definition of Post-Prematurity Respiratory Disease among Extremely Preterm Infants

Publication Number: 530.4321

Colm P. Travers, UAB, Birmingham, AL, United States; Waldemar Carlo, UAB School of Medicine, Birmingham, AL, United States; Amaanti Sridhar, RTI International, Research Triangle Park, NC, United States; Barry S. Eggleston, RTI International, Mebane, NC, United States; Abhik Das, RTI International, North Potomac, MD, United States; Vivek Shukla, University of Alabama at Birmingham, Birmingham, AL, United States; Anna Maria Hibbs, UH Rainbow Babies & Children's Hospital, Cleveland, OH, United States; Matthew M. Laughon, The University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Edward F. Bell, University of Iowa, Iowa City, IA, United States; Matthew Rysavy, McGovern Medical School at the University of Texas Health Science Center at Houston, Houston, TX, United States; Erik A. Jensen, Lucile Packard Children's Hospital Stanford, San Francisco, CA, United States; Brenda Poindexter, Children's Healthcare of Atlanta, Atlanta, GA, United States; Kristi L. Watterberg, University of New Mexico School of Medicine, ALBUQUERQUE, NM, United States; Susan R. Hintz, Stanford University School of Medicine, Palo Alto, CA, United States; Anup Katheria, Sharp Mary Birch Hospital for Women & Newborns, San Diego, CA, United States; Eric Eichenwald, Children’s Hospital of Philadelphia, Philadelphia, PA, United States; Sara B.. DeMauro, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, United States; Namasivayam Ambalavanan, University of Alabama School of Medicine, Birmingham, AL, United States; NICHD Neonatal Research Network, National Institute of Child Health and Human Development, Bethesda, MD, United States

Background: Post-prematurity respiratory disease (PPRD) frequently occurs among survivors of extreme prematurity. Defining the presence and severity of PPRD is important for systematic monitoring of longer-term outcomes.

Objective: To define the presence and severity of PPRD at 18-26 months’ corrected age in preterm infants.

Design/Methods: We included infants < 29 weeks’ gestation inborn at NICHD Neonatal Research Network centers from 2008 to 2019, and who had follow-up at 18-26 months’ corrected age. We excluded infants who died before 36 weeks’ postmenstrual age, or if follow-up was incomplete, or if major malformations or neuromuscular disease. Caregiver questionnaires administered at follow-up collected PPRD relevant data including respiratory support, pulmonary medications, respiratory surgeries, respiratory related rehospitalizations, and pulmonary clinic visits. Various clustering algorithms were evaluated using all PPRD relevant data and fitted with hyperparameter tuning.

Results: 8596 infants were included of whom 395 (4.6%) died before 18-26 months. The Hierarchical Density-Based Spatial Clustering of Applications with Noise algorithm was the best model based on dense and separated clusters with a strong silhouette score of 0.714. The model assigned 7210 infants into eight PPRD clusters. There were 991 (12.1%) surviving infants with an undetermined cluster label. Among all 7605 clustered infants (including deaths), there was no evidence of PPRD in 3073 (40.4%) infants. Grade 1 PPRD (20.1%) included two clusters with one group classified based on few PPRD-related readmissions and one group classified based on discontinuation of all PPRD-related treatments by follow-up. Grade 2 PPRD (30.3%) comprised four clusters identified as more persistent PPRD based on active use of pulmonary medications and/or pulmonary clinics visits, and varying numbers of readmissions. There were 4.0% of infants identified in a cluster with grade 3 PPRD suggested by a combination of ongoing respiratory support (oxygen, CPAP, or ventilation), respiratory surgeries, and a history of respiratory medications and pulmonary clinics. There was 99.6% agreement for presence of any PPRD and 89.6% agreement for the severity of PPRD comparing our cluster-defined PPRD and clinically-defined PPRD based on consensus (Figure 1).

Conclusion(s): Clustering analysis identified data-driven grades of PPRD among infants < 29 weeks based on respiratory illness and treatments by 18-26 month follow-up. The cluster-based definition may serve as a basis for tracking outcome trajectories and the effect of interventions in this high-risk population.

Figure 1
Confusion matrix for Hierarchical Density-Based Spatial Clustering of Applications with Noise results versus clinically-defined results among complete-case survivors. Collapsed PPRD labels defined using the clustering algorithm were graded from 1 to 3 and compared with clinically-defined PPRD based on consensus (none, mild PPRD, moderate PPRD, and severe PPRD).