But before the data are ready for analysis a number of steps have to be taken. All these steps have been documented in detail in our standard operating procedures that are based on International Standards Organization (ISO) 9001:2000 standards. A summary of those steps has been outlined below:

Step 1: Receiving the dataset. Once a year, before November the 1st, all collaborating European national and regional registries are asked to send us their data on the previous year, if they have not already done this in the months before. Data are usually received in encrypted form by email. The sender receives a receipt notification by return.

Step 2: Decryption of the dataset. Because of the encryption, third parties are not able to read or modify the data and the Registry office can be sure that the dataset comes from a trusted party. At receipt the dataset will first be decrypted with the ERA-EDTA Registry key and stored on the file server as a read only file.

During this phase a number of rough checks will be performed. For example, we check whether the data are readable, if they cover the correct time period and whether the national registry provided all required fields. If there is any sign of ambiguity or lack of clarity (e.g. unknown coding systems) we immediately contact the national representative to solve these problems. If there are no problems, the dataset will be forwarded to the data manager for further processing.

Step 3: Import of the dataset into the ERA-EDTA Registry database.

a) Uniformisation of the datasets. As there is a lot of variation in the format in which the datasets are provided (e.g. database structure, variable names, coding systems), the data manager will run a country specific script which contains a number of instructions for uniformisation. Although this method saves a lot of time each year a new dataset has to be imported, it requires that each national or regional registry does not make major changes to the format of their datasets.

b) Translation of local coding systems into ERA-EDTA coding systems. Some registries use their own coding system for the storage of for example primary renal disease or event types. Therefore in this step a conversion is performed by selecting the right cross-map table.

c) Check on the validity of the data. In this step domain checks (e.g. dates in the future?) and several consistency checks (e.g. non-existing patient id's in the transaction table) will be performed. A logging of this procedure and the errors that might occur is stored for future reference.

d) Import of the dataset in the database. At this moment the dataset will actually be imported in the ERA-EDTA Registry database which is implemented in MS SQL server 2000. Although the data passed the validity exam, there is no proof yet that these newly imported data yield valid statistical results. Therefore the data will first be imported in an acceptance database and not in the production database.

Step 4: Production and submission of a standard report.

Based on the data in the acceptance database, several statistics (e.g. mean age, number of new patients and number of patients in stock) will be produced and checked with other reports (e.g. previous ERA-EDTA reports and national registry reports) on their validity. These statistics will be submitted together with the import report (see below) to the national registry representative with the request to check whether they truly represent a correct interpretation of their data.

Step 5: Happy end.

In case the national representative confirms that the statistics produced by the ERA-EDTA Registry office are valid, the data will be imported in the production database and become permanent. After all datasets have passed through this process the production database is ready and can be used for the production of a new volume of the annual report and for scientific papers.

Vianda Stel, epidemiologist

Paul van Dijk, medical information scientist