Control for bias |

1. Defined sample |

2. Representative sample |

3. Blind outcome rating |

Appropriate measurement of variable |

4. Methods of assessment (a) face validity |

5. Methods of assessment (b) evidence of psychometrics |

6. Outcome data reported |

7. Adequate power calculation |

Control for confounding |

8. Statistical adjustment: multivariable or regression analysis conducted, with adjustment for potentially confounding factors. |

Each of these criterions was explicitly judged by using: Yes = criterion clearly satisfied; No = criterion clearly not satisfied or unclear if criterion is satisfied. A quality score (maximum score = 8) was allocated to each individual study. |