A national internet-linked based database for pediatric interstitial lung diseases: the French network

The RespiRare® database structure for rare respiratory diseases was created to include extensive information on all forms of rare lung diseases grouped within four main categories: ILD, respiratory malformations, ciliary dyskinesia, and other rare diseases with chronic respiratory insufficiency. Data from patients is entered via a secured Internet protocol into a safe database through a web interface. The interface is visible from the web through the https secure protocol, on a DMZ network. It uses the LAMP system, with a MySQL database for data storage; PHP scripts interface for access to database, as well as a Apache 2 webserver. The structure is hosted on Linux servers, within the secure server network of Paris-6 University Pierre et Marie Curie (UPMC), the coordinator affiliated institution. The database server hosting the MySQL server is behind a firewall, linked to the webserver through a secure encrypted IP to IP tunnel. Access to the patient is filtered through a central identification number. Database and data collection have been approved by the French national data protection authorities: “Commission Nationale de l’Informatique et des Libertés” and the “Comité Consultatif sur le Traitement de l'Information en matière de Recherche dans le domaine de la Santé”.

The Reference centre and 8 affiliated centres in university hospitals (called Competence Centres) are spread all around the French national territory. Each of them covers a French area with a local network of affiliated centres (Additional file 1). Every participating centre only has access to their own patients’ data. For cross-centre studies, the respective centres must agree explicitly to share its anonymous data with other centres. A charter describes the rules, including general rules relating to organization and rules governing access to data. Each participating member is asked to accept the charter of good use. To access the database, users receive a personal user name and password. Users can see who modified the file, and if any modification is observed in data recording, a tracking system is available to get back to a previous version of the data. Patients must give their oral informed consent before their data is included into the database.

The quality of the data relies on the users who enter the data, and on a database scientific committee who regularly reviews proposed cases for inclusion. The RespiRare® administration offers each centre a specific training for proper use of the on-line database. A data manager and a technical team are in charge of quality control, monitoring for data coherence, absence of duplicates, and transfer of data if needed. The data is organized in a data warehouse to optimize data visualization and data retrieval.

To maintain motivation and to ensure long-term involvement of all participants, specific contacts with the coordinating team through e-mail, phone calls and local visits have been put in place. In addition, an annual meeting is organized with all affiliated centres.

The database we have created at the national level in France was elaborated by a team composed of clinicians and databank professionals, using data modelling techniques and methodologies with a Conceptual Data Model (CDM) providing a view of the key data entities and their relationships (Additional file 2). The RespiRare® database is structured as medical record for patients. Users can document a large amount of data: identity, social data, environmental data, and diagnosis of the ILD if known, genetic data, patients visits to the centre, and the entire medical examinations and tests done for the diagnosis or the follow up of the patient. Data can be entered at every visit of the patient or at least once a year in a minimal dataset containing an “annual form” and a “diagnosis form”, which is filled at the first visit and can be updated at any time (Additional file 3). The datasets for the 3 other disease groups (respiratory malformations, ciliary dyskinesia, and other rare diseases with chronic respiratory insufficiency) are organized in a similar way, with a diagnosis dataset and a follow-up dataset (on an annual basis). A common list of items for all groups includes clinical, functional, imaging, organ specific tests, treatments and outcome information. Extended disease-specific datasets contain information relevant to each group of rare diseases. The entire ILD French database dictionary is provided in the Additional file 4. These extended datasets have been defined by steering committees composed of experts in the respective groups of diseases. They are implemented continuously in order to provide extensive and detailed documentation of a given disease. The database also offers a system for inclusion of biological, microbiological, radiographic, histological, cytogenetic, molecular and genetic information.

The database is designed for a long-term use. In this view, the interface has been developed so that it is easy to use and friendly, with rolling menus and data entry forms accessible to unskilled users. In addition, RespiRare® database has been organized and structured as a medical record for patients. Consequently, physicians can use the system as a daily aid for patient care through a web accessible medical record backed on this database. It takes around 20 minutes to fill the form at the first visit and around 5 minutes at every additional visit.

Based on our previously described ILD classification (Clement et al. Orphan J Rar Dis 2010), all patients eligible to one of the ILD diagnosis can be proposed for inclusion in the database (Table 1). Moreover, if no diagnosis is known, any patient under 18 years old with a persistent interstitial pathway on CT-scan can also be included in the database. Acute interstitial diseases are not included. The diagnosis is based on clinical, radiological, and functional features, including tachypnea, diffuse infiltrates on chest imaging and impaired gas exchange. Written information is given to the patient/family before documenting the database. The physician of the Competence Centre in charge can recommend that patients be sent to the Reference Centre for an exam or a specific investigation.

The RespiRare® database was launched in September 2008. The database was set up at the end of 2008. After their initial training, centres started to include patients one by one. At the beginning, after the database committee validation, centres included all patients with ILD less than 18 years old currently followed in the centres, with their entire past medical records. New patients were also included in the database in a prospective way. Thus, the database covers up to 17 years for some patients followed for a long time at the time of the database creation. This is an on-going process, and centres’ ability to propose patients for inclusion in the database is still variable in terms of speed. Multiple biases, such as recovered or deceased patients, cannot let us talk yet about prevalence or incidence.

At the present time, 217 patients were proposed for inclusion in the database. The database committee rejected 12 patients because the diagnosis of ILD wasn’t proven, or because insufficient data was provided. Thus, the number of ILD patients included in the database is now 205.

The study is presented to all patients/families during a medical visit. None of the family has refused to be enrolled. Of importance also is the observation that many physicians in charge of the patients consider this structured data collection of great help. Indeed, as mentioned above, in an increasing number of French pediatric pulmonology teams, RespiRare® database is now used as a medical record for patient care.

Considering the 205 ILD children entered in the database, the sex ratio is 0.9 boy/girl. The median age at diagnosis is 1.5 years old [0–16.9]. More than half of ILD start before 2 years old (Figure 1). Median diagnostic delay is 1 year [0–12.6], largely influenced by the type of ILD. The median length of follow up into the database is currently 2.9 years [0–17.2].

Aetiology for the ILD has been identified for 149 (72.7%) patients whereas 56 (27.3%) patients remain with no identified cause. Surfactant deficiencies and alveolar proteinosis, haemosiderosis and sarcoidosis, represent almost half of the diagnoses. The diagnosis breakdown, age at diagnosis, sex ratio and length of follow up are detailed on Table 2.

For 36 children (17.5%), a surfactant disorder has been identified. These children have an early median age of diagnosis (0.3 year-old). Pulmonary surfactant-associated protein C gene (SFTPC), Pulmonary surfactant-associated protein B gene (SFTPB), ATP-binding cassette subfamily A, member 3 gene (ABCA3), and Thyroid transcription factor-1 gene (TITF1 = NKX2-1) mutations were found (Table 3).

SFTPC mutations are the most frequent genetic abnormalities (22 patients), representing 10.7% of all ILD patients. The recurrent mutation I73T represents two thirds of them. The median age at diagnosis is 6 months [0–8.5]. Lung biopsy was performed for 14 children, showing an alveolar wall thickening, type 2 alveolar epithelial cells hyperplasia, inflammation cells infiltration with or without macrophagic intra-alveolar accumulation with cholesterol crystals and occasionally fibrosis. All patients received steroids. Among them, nineteen patients underwent intravenous steroid pulses and azithromycin. Two patients died at 3 and 17 years old. Seven patients presenting neonatal respiratory distress had SFTPB homozygous mutations. All of them died before 2 months of life. ABCA3 homozygous mutation was found in 5 patients with neonatal respiratory failure. Median age at diagnosis was 0.3 years old [0-12.3]. They all needed long-term steroid pulses. Two patients showed a brain lung thyroid syndrome with TITF1/NKX2-1 mutation. One of them died at month 18.

Twenty patients were diagnosed with alveolar proteinosis (sex ratio = 1.9 boy/girl). Onset of symptoms was very early with a median age at diagnosis of 7 months. Most of them needed therapeutic broncho-alveolar lavages (BAL).

We collected data of 23 haemosiderosis (Table 4). There was a large predominance of girls (18 vs. 5 boys) and the median age at diagnostic was 4.8 years old [0.7-14]. Three of them were associated with another disease (cow proteins milk allergy, celiac disease, Down syndrome). Initial presentation was mainly acute with haemoptysis and major anaemia but some cases were more difficult to diagnose with cryptic chronic bleeding and no haemoptysis. Pulmonary hypertension was reported for only 2 children. Nine children underwent open lung biopsy or transbronchial biopsy to confirm the diagnosis. All children benefits from a steroid treatment, with steroid pulses for 11 of them. Four patients benefit from supplemental treatment by hydroxychloroquin and 4 from mycophenolate mofetil.

Twenty-one children had interstitial granulomatous lung disease, which was for all of them a sarcoidosis (Table 5). Sex ratio was 1 boy/girl and median age at diagnosis was 9.4 years old. Sarcoidosis is one the most frequent pre-teenager ILD diagnosis with connective lung diseases and exposure related ILD. BAL was performed for 15 of them and showed in most of the cases a lymphocytic infiltration. CD4/CD8 ratio was available for 7 patients and was >1 for 6 of them. Five patients underwent transbronchial or open lung biopsy to assert diagnosis. All patients were treated with steroids and among them, 11 received steroid pulses.

Nine children had exposure related ILD: 5 had hypersensitivity pneumonitis (turtledove, pigeon, straw), 1 was due to total body irradiation, and the causal agent was not specified for the others. Outcomes were good with no residual symptoms after eviction and, for some of the patients, short steroid therapy.

Connective lung diseases were diagnosed for 9 children (rheumatoid arthritis, systemic sclerosis, mixed connective tissue disease). Three children presented with vasculitis of unknown origin. No capillaritis has been included in the database at this time. All together, there are twice as many girls than boys and teenagers are largely represented, but 3 children were diagnosed before 5 years old.

ILD related with metabolic disorders were diagnosed for 5 children. Among them, 4 had a Niemann-Pick B or C disease and one had a Cobalamine C deficiency.

Five children had Langerhans’ cell histiocytosis with ILD. All but one had a multifocal disease with bone involvement.

We also collected data for 4 eosinophilic lung diseases and 4 lymphatic disorders.

Nine children presented an ILD related to others diagnosis: Crohn disease (n = 1), inhalations (n = 1), alpha-1-antitrypsin deficiency (n = 1), xanthogranulomatous disease (n = 1), constrictive bronchiolitis (n = 1), hamartoma (n = 2), others (n = 2).

Finally, out of 205 patients presenting with ILD, 56 children (27.3%) remained with no diagnosis (Additional file 5). Sex ratio is 1 male/female and median age at diagnosis is young: 0.7 years [0–16.4]. Of importance is the observation that 34.5% of children with ILD diagnosed before 2 years old remain with no diagnosis whereas it is so for 19.1% of ILD diagnosed after 2 years old. For a large part of them, the absence of diagnosis was concluded after a number of investigations, including genetic studies (n = 40) and BAL (n = 27). Only 22 of them underwent lung biopsy. However, some patients did not benefit from these explorations, or are missing data, so that the absence of diagnosis cannot be ascertained.