Assessing the quality of reports of randomized clinical trials: Is blinding necessary?