Introduction
Childhood interstitial lung disease (chILD), also referred to as diffuse lung disease (DLD), is a heterogenous group of disorders with overlapping clinical, radiographic, and pathologic characteristics. While chILD is considered rare, the incidence and prevalence are not well established.1 Over the past 30 years, the scope of the field has expanded through recognition of specific disease entities. For example, surfactant protein B deficiency was first described in the early 1990’s,2 with reports ofSFTPC and ABCA3 mutations as causes of chILD in 2001 and 2005, respectively,3; 4 followed by the definite description of Neuroendocrine cell Hyperplasia of Infancy (NEHI) in 2005.5 After development of consensus terminology and lung biopsy classification, the first clinical practice guideline was published in 2013.6-8 Further advances in molecular genetics and imaging technology have subsequently modified the diagnostic approach, though limited data are available on the clinical utility of genetic testing for chILD.
In 2012, a revision to the International Classification of Diseases (ICD) Ninth Revision added codes for several chILD disorders under an umbrella diagnosis of “Other alveolar and parietoalveolar pneumonopathy”, which included code for surfactant mutations of the lung, NEHI, and other interstitial lung diseases of childhood.9 The currently utilized tenth revision of ICD codes expanded the codes for systemic disease associated with pulmonary manifestations10. Billing codes have not previously been utilized as a case ascertainment method for research in chILD.
In this study, we sought to evaluate recent practices at a tertiary care medical center to better understand chILD prevalence, care patterns, healthcare utilization, and outcomes. We identified diagnostic challenges and found substantial inpatient and multidisciplinary outpatient care utilization. By highlighting the scope of this patient population and associated care needs, this knowledge can be leveraged to inform healthcare delivery in the future.
Methods 
Study population
A single-center, retrospective cohort study of patients receiving care at the Children’s Hospital of Philadelphia (CHOP) was performed. This study was granted exempt approval with waiver of informed consent from the CHOP Institutional Review Board (IRB 19-016269). Patients were identified based on a query of selected ICD-10 billing codes (Supplemental Table 1 ) relevant for chILD/DLD. Medical records were then reviewed for patients who had either outpatient or inpatient pulmonary encounters between January 1, 2019, and December 31, 2021, curating cases for inclusion after chart review to confirm ILD/DLD diagnosis. Data collection included EMR data through September 28, 2022, allowing time for diagnostic testing and clinical follow up after initial evaluations.
Study design and data collection
Outpatient and inpatient encounters, medication exposure data, and procedural history were obtained via electronic medical record (EMR) reports developed for this study. Supplemental chart review was performed, focused on assessment of diagnostic testing performed outside of CHOP and characterization of disease severity and outcomes. Study data were collected and managed using Research Electronic Data Capture electronic data capture tools hosted at CHOP.11; 12Disease categories or classification were designated by the study team based on the aggregate of available clinical data, genetic testing, imaging, and lung biopsy reports in the EMR.
Statistical analysis
We analyzed EMR data based on lifetime to date and the time period between 2019-2021. Descriptive statistics were utilized, with reporting of means and standard deviations (SD) for continuous variables and frequencies and percentages for categorical variables.
Results Patient demographics and clinical characteristics
Through ICD-10 billing diagnoses, a total of 306 patients were identified as receiving care during this 3-year period. Demographic characteristics are summarized in Table 1 . A slight majority (55%) of patients were male, and 77.1% were born as a result of full-term (≥ 37 weeks) pregnancies. More than half (56.2%) of the cohort identified as white race and two-thirds (66.7%) had private insurance, in comparison to white race for 53.3% and private insurance for 52.4% among all patients seen with outpatient encounters in the Division of Pulmonary and Sleep at CHOP during the same time period. For those patients with a documented age of onset of respiratory symptoms (n=231), 40.7% had respiratory symptom onset in the first two years of life, with 30.3% developing symptoms before six months of age. The onset of respiratory symptoms was not documented in the EMR for 24.5% of the cohort. Overall, 70 (22.9%) patients were referred for a second opinion after seeing a pulmonologist at another institution.
Distribution of ILD diagnoses
DLD associated with oncologic disorders was the most common diagnostic category, accounting for 65 (21.2%) cases (Table 2 ). We also identified cases of bronchiolitis obliterans (n=31, 10.1%), DLD associated with connective tissue diseases (n=29, 9.5%), NEHI (n=28, 9.2%), environmental/toxic related lung injury (n=27, 8.8%), and alveolar hemorrhage disorders (n=23, 7.5%). Surfactant metabolic dysfunction was responsible for 19 cases (6.2%), while 15 cases (4.9%) had DLD associated with immune-related disorders. There were multiple other diagnoses for which three or fewer patients were identified (Supplemental Table 2 ). For 32 cases (10.5% of the cohort), a definitive diagnosis or classification could not be assigned based on EMR data. The overall cohort mean age at time of diagnosis was 7.3 years (SD 6 years).
Diagnostic testing and monitoring
Almost all patients (92.2%) had chest computed tomography (CT) scans, with 220 (71.9%) patients having at least one chest CT between 2019-2021. Flexible bronchoscopy was performed in 97 (31.7%) patients ever in their lifetime, with procedures in 73 (23.9%) patients during the recent 3-year period. Echocardiography was performed in 195 patients (63.7%) ever in their lifetime and half (n=153, 50%) of patients during the 3-year period, with a mean of 2.1 (SD 2.7) procedures per patient per year.
Lung tissue sampling was performed in 110 (35.9%) patients in a variety of contexts (Figure 1A ), with 72 (65.5%) patients having the procedure at our center at a mean age of 7.6 years (SD 6.3 years). Fifty-two (72.2%) patients had a wedge biopsy. In addition, there were 13 transbronchial biopsies, six lung nodule biopsies, three endobronchial ultrasound with transbronchial needle aspiration procedures, two lobectomies, two bronchial biopsies, two core needle biopsies, and one endobronchial mass excisional biopsy. Of the 23 patients identified with alveolar hemorrhage, most (n=20, 87%) underwent lung biopsy, resulting in histologic diagnosis of capillaritis in 50%. Over half (52.6%) of patients ultimately diagnosed with surfactant metabolic dysfunction underwent lung biopsy. Notably, 45 (40.1%) patients had a lung biopsy performed at an outside institution; seven of these patients subsequently had a lung biopsy repeated at our center.
In total, 149 (48.7%) patients underwent genetic testing relevant for their lung disease evaluation (Figure 1B ), and 38 (25.5%) of tested patients had a diagnosis established by genetic testing (Figure 2). Alterations in genes impacting surfactant metabolism dysfunction accounted for 50% of positive results. CHOP providers ordered the testing for 22 (57.9%) of positive results. The 76-gene CHOP Comprehensive Pulmonary-Vascular Panel was performed for 41 cases and yielded positive results in four cases. Two patients were found to be compound heterozygous for variants in ABCA3 , one patient was found to be heterozygous for a de novo pathogenic variant in SFTPC , and one patient was heterozygous for a pathogenic deletion within chromosome 14q12, which includes the NKX2.1 gene. CHOP providers ordered other respiratory relevant genetic analyses in 61 patients, including single gene testing/targeted variant testing and targeted pulmonary gene panels for 47 patients, 14 of which resulted in a molecular diagnosis. Chromosomal microarray analyses were performed in 21 patients, all of which were non-diagnostic. Respiratory-relevant genetic testing was performed prior to referral to our center in 41 patients, with molecular diagnosis established by single gene testing/targeted variant testing and targeted pulmonary gene panels for 14 (37.8%) of 37 patients.
Whole exome sequencing (WES) was performed in 58 (38.9%) patients, was ordered by CHOP providers in 42 (72.4%) instances and was diagnostic for six (10.3%) patients. Positive findings resulted in diagnoses of CHOPS syndrome (n=2), ataxia telangiectasia (n=2), lysosomal storage disease (n=1), and one diagnosis of an immune-related disorder (STAT1 gain-of-function variant). Two of the patients diagnosed through WES also had non-diagnostic targeted pulmonary gene panels, including one patient with CHOPS syndrome and one with lysosomal storage disease (GM1 gangliosidosis).
In examining the diagnostic categories of patients who underwent genetic testing within the overall cohort, NEHI made up the largest group (n=21, 14.1%), followed by those with ILD not otherwise specified (NOS) (n=19, 12.8%), surfactant metabolic dysfunction (n=18, 12.1%), DLD associated with connective tissue disorders (n=16, 10.7%), and bronchiolitis obliterans (n=14, 9.4%). Patients with surfactant metabolic dysfunction and those ILD NOS were among those most likely to have undergone genetic testing (94.7% and 82.6%, respectively, Figure 1B ).
Both genetic testing and lung biopsy were performed in 63 (20.6%) unique patients. Those with alveolar hemorrhage disorders were the largest subgroup (n=12, 19.0%), followed by nine (14.3%) patients with DLD associated with connective tissue disease and seven (11.1%) patients with surfactant metabolic dysfunction. We analyzed the time order of testing for the 20 patients who had the CHOP Comprehensive Pulmonary-Vascular Panel and lung biopsy performed, as we could confirm the order timing within our institution. Eleven patients had lung biopsy performed prior to genetic testing, while eight had genetic testing performed first. One patient had lung biopsy performed on the same day that genetic testing was initiated. Ultimately, six patients who had a lung biopsy later had a genetic diagnosis established.
Outpatient care utilization
Within this cohort, nearly all patients (96.1%) received outpatient pulmonary care at our center. A mean of 2.8 (SD 2.6) outpatient pulmonary visits per patient per year occurred between 2019-2021. Cardiology, endocrinology, and gastroenterology were other highly utilized medical subspecialties (Figure 3 ). At our center, more patients had encounters with nutrition and physical therapy than with speech and language pathology or occupational therapy.
Hospitalizations
We quantified the hospitalization events using EMR data for hospital days at CHOP both for lifetime to date and for the 3-year period from 2019-2021. In their lifetime, 246 (80.4%) patients had ever been hospitalized at CHOP, while nearly half (n=138, 45.1%) had ever been hospitalized in an intensive care unit (ICU). Between 2019-2021, 176 (57.5%) patients were hospitalized, with a mean of 42.1 (SD 80.6) hospital days and 3.1 (SD 3.5) hospitalizations per hospitalized patient. ICU admission occurred in 82 patients, reflecting 26.8% of all chILD patients and 46.6% of the hospitalized chILD patients, and accounting for a mean of 38.4 (SD 93.2) ICU days and 1.8 (SD 1.9) ICU hospitalizations per ICU hospitalized patient. When considering hospitalization burden across the entire cohort, the mean number of hospitalization events per patient was 1.8 (SD 3.0) during this 3-year period, with 0.5 (SD 1.3) ICU hospitalization events per patient.
In examining hospitalization data based on diagnostic category from 2019-2021, the largest group of hospitalized patients was DLD associated with oncologic disorders (n=34, 19.3%). Other groups of patients who accounted for the most hospitalizations were patients with environmental/toxic related lung disease (n=26, 14.8%), bronchiolitis obliterans (n=16, 9.1%), alveolar hemorrhage disorders (n=16, 9.1%), and DLD associated with connective tissue disease (n=19, 10.8%). Within each of these disease categories, most patients were hospitalized at least once in their lifetime (Figure 4A ). Between 2019-2021, the percentage of ICU hospitalization by disease category was greatest for patients with surfactant metabolic dysfunction (42.1%), alveolar hemorrhage disorders (30.4%), and DLD associated with oncologic disorders (27.7%) (Figure 4B ).
Medication utilization
Medication exposure most relevant for treatment of chILD/DLD was examined. Due to the EMR driven, retrospective methodology of our study, medication exposure was defined as documentation of an inpatient order or presence of medication on the medication list associated with an outpatient encounter. In examining lifetime exposure, 78% of patients had some form of oral and/or intravenous corticosteroid ever prescribed, and 57% were prescribed intravenous corticosteroids (methylprednisolone). Between the years of 2019 and 2021, 10.5% of patients were prescribed intravenous corticosteroids, while 29.4% had oral and/or intravenous corticosteroids prescribed.
Hydroxychloroquine was prescribed for 10.5% (n=32) of the overall cohort across their lifetime, with 7.8% (n=24) of the cohort receiving hydroxychloroquine between 2019 and 2021 at a mean age of 10.6 years (SD 6.4) at the time of the first prescription. The most frequent underlying diagnoses for patients prescribed hydroxychloroquine were surfactant metabolic dysfunction (n=11, 45.8%) and DLD associated with connective tissue disease (n=7, 29.2%). Pirfenidone was prescribed for eight (2.6%) patients, including three with DLD associated with oncologic disease, two with DLD associated with connective tissue disease, and one each with bronchiolitis obliterans, surfactant metabolic dysfunction, and fibrotic lung disease of unknown etiology.
Respiratory support and Outcomes
In the overall cohort, 32.7% had ever required chronic home supplemental oxygen, 12.4% had ever received chronic non-invasive respiratory support, and 4.6% underwent tracheostomy for chronic invasive respiratory support. The mean age at time of tracheostomy placement was 4.8 (SD 6.1) years. Chronic supplemental oxygen use was prevalent in patients with NEHI, accounting for 24% of the group with home supplemental oxygen use in general.
Of the 306 patients seen between 2019-2021, seven (2.2%) patients had died as of January 1, 2022, including four with DLD associated with oncologic disorders, one with DLD associated with connective tissue disease (juvenile dermatomyositis with pulmonary involvement), one with surfactant metabolic dysfunction (SFTPC ), and one with DLD associated with immune-related disorders (severe combined immunodeficiency disease). Five deaths were to due respiratory failure, while in two patients cause of death was not documented. Pulmonary fibrosis was associated with mortality in four patients, including the one with connective tissue disease (juvenile dermatomyositis), one with ILD due to SFTPC , and two with prior history of oncologic disease. Further, thirteen patients underwent lung transplantation including four patients with surfactant metabolic dysfunction, three with DLD associated with oncologic disorders, two with vascular disease not otherwise specified, and one each with alveolar capillary dysplasia, bronchiolitis obliterans, pulmonary capillary hemangiomatosis, and ILD of unknown etiology. The four patients with surfactant metabolic dysfunction who underwent lung transplant included two patients heterozygous for variants in ABCA3 , and two patients with surfactant protein C associated ILD. Two other patients were listed for lung transplant, and one was awaiting combined heart/lung transplant.