Article type
Year
Abstract
Objective: The benefits from thrombolytic therapy in patients with ischaemic stroke are uncertain. We decided to perform a "prospective" individual patient meta-analysis in an attempt to identify patients who could benefit from treatment and those who have a higher risk of haemorrhagic transformation. Here we will present the problems encountered.
Methods: A Steering Committee composed of representatives from participating trials wrote the protocol, defined a core of variables to be entered into the cornmon file, and agreed on the access rules to the database. This was achieved before initiation or while the trials were being run. Data were provided upon completion of the trials.
Results: Major problems: The problems can be divided into two types, nonspecific i.e. related to the approach used and specific, i.e. related to the clinical problem. The nonspecific problems included the data release policy, because it is an awesome task to convince trialists to pool their data, even when their data were generated using public funds. When data belonged to pharmaceutical industries, claiming for cooperation was the only possible action but was unsuccessful. Only four of the six trials accepted to pool their patients' data in the TASPP common file. The specific problems included the need to generate some variables; to ensure a good homogeneity, each variable had to be checked, and the data compared between trials. Also different neurological scales were used for patient assessment at baseline, making it difficult to find a common classification for severity at entry, a major prognostic variable. There was no common definition for early CT-scan abnormalities, although all the trials had similar central CT-scan assessment procedures.
Discussion: The technical problems could be solved through working sessions which were time consuming, but the "political" obstacles could not be overcome. Ways of enabling easier access to data for researchers, while conserving confidentiality and other characteristics is a challenging issue which should be addressed in the near future.
Methods: A Steering Committee composed of representatives from participating trials wrote the protocol, defined a core of variables to be entered into the cornmon file, and agreed on the access rules to the database. This was achieved before initiation or while the trials were being run. Data were provided upon completion of the trials.
Results: Major problems: The problems can be divided into two types, nonspecific i.e. related to the approach used and specific, i.e. related to the clinical problem. The nonspecific problems included the data release policy, because it is an awesome task to convince trialists to pool their data, even when their data were generated using public funds. When data belonged to pharmaceutical industries, claiming for cooperation was the only possible action but was unsuccessful. Only four of the six trials accepted to pool their patients' data in the TASPP common file. The specific problems included the need to generate some variables; to ensure a good homogeneity, each variable had to be checked, and the data compared between trials. Also different neurological scales were used for patient assessment at baseline, making it difficult to find a common classification for severity at entry, a major prognostic variable. There was no common definition for early CT-scan abnormalities, although all the trials had similar central CT-scan assessment procedures.
Discussion: The technical problems could be solved through working sessions which were time consuming, but the "political" obstacles could not be overcome. Ways of enabling easier access to data for researchers, while conserving confidentiality and other characteristics is a challenging issue which should be addressed in the near future.