Guidelines, meta-analysis and reviews conclude on and recommend an intervention for stroke patients if there is evidence for its effectiveness. Physicians prescribe those interventions if only these written sources recommend them. Worse than that, insurance companies pay the bill according to these reports. However, nobody discuss the generalizability of findings in these written sources. Nobody knows in what percent the study population of these published data represent the whole stroke patients in rehabilitation settings (in- or outpatient based). No need to mention that we already excluded the patients who were dead before rehabilitation stage.

There could be some stroke patients in somewhere on the world that we can never learn their outcome, as they had no chance to reach a health professional who has the ability to follow these patients and perform standardized outcome assessments. Or there can be a lack of a researcher in the patient's institute who will enter these data to a database, do statistics (good enough to pass the editor's basic standards) and able to write the results in only couple worldwide known languages, send the manuscript and never give up waiting for the answer with endless patience. Everything may go perfect up to here, but if the journal which accepted this manuscript to publish is not indexed in a common database, clinical findings still may not turn into evidence.

In some developing countries, patients are loosing their treatment rights as insurance companies decide not to cover some physical modalities. These thirdparty payers based their cancellations to western research which were performed on a different patient population at a different rehabilitation setting. In a comprehensive rehabilitation setting with well performed treatment techniques by well-trained rehabilitation team members the so called experimental intervention may not be found superior to standard treatment however it may mean a lot in a developing country with limited resources.

It is not possible to smooth every relevant error source in research but we must at least know for which stroke subpopulation the evidence is applicable. In clinical trials, many stroke patients are excluded due to several reasons (i.e. cognitive impairments, contraindications to the experimental intervention etc). The inclusion and exclusion criteria are mostly presented in detail under methodology part. Studies should be planned to document the inclusion and exclusion criteria and target population of published articles presenting the effectiveness of an intervention for stroke patients as well as to find out the specific stroke subpopulation that the findings of written resources can only be applicable for. Studies reporting the bias caused by patients, health-professionals, researchers and socio-economical status are needed.