Discussion
In this retrospective single center study, we used ICD-10 billing diagnoses to identify patients with chILD/DLD. This strategy allowed us to identify 306 children and adolescents with chILD/DLD who received care at a large volume tertiary care medical center between 2019 and 2021, providing new data on the scope of chILD. Furthermore, using EMR data collection, we provide the first report on health care utilization and hospitalizations in chILD in the United States (U.S.).
The diagnosis of chILD/DLD is challenging and there is uncertainty about incidence and prevalence. Published estimates are 0.13-16.2 cases per 100,000 children per year,1 acknowledging reports are based on differing case definitions and study methodology. For example, the 2004 European Respiratory Society task force identified 185 cases of interstitial lung disease in children over a five-year period13 and a 2018 study by the chILD European Union registry identified 575 patients across 82 centers from 16 countries over a three period.14 In the U.S., the chILD Research Network reported 187 lung biopsy cases in children <2 years of age from 11 centers from 1999-2004 and 191 cases in children ages 2-18 from 12 centers over a four-year period.6; 7Additionally, Soares et al. performed a retrospective review at a mid-size single center, identifying 93 cases over an 18-year period.15 Our study methodology did not differentiate incident cases from prevalent cases. Our cohort size may reflect a combination of the overall size of our children’s hospital, the study methodology, and increased awareness and clinical suspicion for chILD in recent years. A substantial portion of cases were seen by pulmonologists at other centers prior to referral to our center, further limiting any population-based estimates and volume predictions for other pediatric centers. We do note that the size of our identified chILD cohort is very similar to the number of patients seen at our Cystic Fibrosis Center between 2019 and 2021.
The distribution of diagnoses of identified patients is also instructive in considering the spectrum of chILD and the care delivery models. Our chILD cohort highlights the large number of patients with systemic diseases with lung involvement or pulmonary complications, particularly those with history of oncologic, connective tissue, and immune-mediated diseases. Immunocompromised patients accounted for approximately one quarter of chILD cases. Disorders that typically present in infancy, including NEHI, surfactant metabolic dysfunction, and diffuse developmental lung diseases were less prominent in our cohort than in prior lung-biopsy based studies. Reflecting the time period of our study, we found a notable group of adolescent patients with e-cigarette or vaping associated lung injury (EVALI), with either abnormalities on chest CT or lung function. However, based on our clinical experience, we suspect that the billing query approach underestimated the number of patients receiving care for EVALI during this time.
We were particularly interested in examining the utilization and impact of genetic testing in our cohort, as many prior studies have relied heavily on lung biopsy classification. Although 51% of the cohort had respiratory disease directed genetic testing, a molecular diagnosis was identified in only 38 cases, reflecting 25.5% of those tested and 12.4% of the overall cohort. Despite advances in understanding the genetic underpinnings of chILD, this data suggests opportunity for further discovery and application of newer genomic technologies in a more comprehensive manner. In addition to limitations in diagnostic yield, barriers to testing may include cost and/or insurance coverage, other access barriers, and turn-around-time for acutely ill patients.14; 16-18 The potential value of negative genetic testing results should also be acknowledged and further examined. While a subset of patients at our center had both lung biopsies and genetic testing, further prospective studies are needed to examine how both positive and negative genetic results impact clinical care decisions.
In addition to ambulatory medical care utilization, chILD patients were frequently hospitalized, including in the ICU. While oncology patients were the largest group, and our data do not allow us to delineate whether the hospitalization was primarily attributable to lung disease, patients with a broad representation of chILD etiologies were hospitalized across the age spectrum. Comprehensive data were not available on hospitalization and testing performed outside our institution, though the frequency of second opinion referrals and outside lung biopsies suggests that our data underestimate the overall healthcare utilization of this cohort. A recently published prospective, longitudinal study by Seidel et al. evaluated healthcare utilization and medical costs in chILD among 445 patients in 10 European countries, reporting a high health care economic burden.19 The authors found that outpatient care, inpatient hospital admission rates and inpatient hospital days were high, though decreased over time as follow up was performed upwards of five years after baseline data was obtained. Medical costs were reported to be highest in those with diffuse developmental disorders and diffuse parenchymal lung disease of unclear etiology in the non-neonate. While our study does not directly consider the medical cost of chILD, our data demonstrate the impact to patients, families, and the healthcare system in the U.S.
Our study highlights the power of EMR-derived data but also identified some opportunities and challenges. First, historical data are limited for older patients, as the current EMR system was implemented about 10 years ago. Because hospitalization billing diagnoses are assigned administratively and not by medical providers at our center, chart review would be required to determine whether lung disease was a primary or secondary indication for hospitalization events in many cases. ILD diagnosis benefits from multidisciplinary case review, and we suggest development of standardized approaches to document the outcome of such reviews, as well as other discrete data elements in the EMR that inform individual patient care and population management. For example, although we know that the age of symptom onset is important for the differential diagnosis in a patient with suspected chILD, this information was not documented in the EMR for 24.5% of our cohort. Lastly, there are limitations in using current ICD billing codes for the purposes of case ascertainment, due to inconsistent use and incomplete correlation with disease categorization in some cases.
In summary, this single tertiary-care referral center report elucidates key aspects of the scope of clinical care and healthcare utilization in chILD, including limitation in current genetic testing and diagnostic capabilities. Further, we demonstrate that use of an EMR approach has the potential to provide new opportunities for cohort analysis that will inform the feasibility of future studies in chILD. Better understanding of these disease processes and mechanisms to monitor patients are needed to advance clinical care delivery and improve outcomes.