The following methodological issues were considered when assessing studies for inclusion: selection, confounding, and measurement. Study designs are often graded hierarchically according to their quality, or degree to which they are susceptible to bias. The hierarchy indicates which studies should be given most weight in a synthesis. In this review, the degree to which each study dealt with the methodological issues was graded into three levels of evidence:
LEVEL A (HIGHEST QUALITY OF EVIDENCE, MINIMAL RISK OF BIAS)
• Prospective studies that started within one year of either initiation or discontinuation of water fluoridation and have a follow up of at least two years for positive effects and at least five years for negative effects.
• Studies either randomised or address at least three possible confounding factors and adjust for these in the analysis where appropriate.
• Studies where fluoridation status of participants is unknown to those assessing outcomes.
...
Studies meeting two of the three criteria for a given evidence level were assigned the next level down. For example, if a study met the criteria for prospective design and blinding for level A, but was neither randomised nor controlled for three or more potential confounding factors, it was assigned level B. Evidence rated below level B was not considered in our assessment of positive effects. However, this restricted assessment of the evidence for Objective 3, so the best level of evidence relevant to this objective (from any study design) was included. In our assessment of possible negative effects, all levels of evidence were considered. Adjustment for confounding factors required analysis of data, simply stating that two study groups were similar on noted confounding factors was not considered adequate.