Table 4

Quality of evidence assessment tool. (1) Is the study, a randomised, double-blind, placebo- (or sham-) controlled clinical trial? Yes / no. (2) considering the design of the trial, please answer the following questions:

Question to respondentInvestigator or sponsor responseReviewer responseSupporting evidence statement
In your opinion, is the study associated with the following types of bias?5-option response (strongly agree–strongly disagree) or not applicable5-option response (strongly agree–strongly disagree) or not applicable
3Comparison group
4Fraud and/or misconduct
5Funding availability
6Hidden agenda
12Sample choice
14Selective reporting
16Wrong design
  • Quality of evidence score = ________% .If the response to question 1 is ’yes’, then the initial quality of evidence score is 100%. If the response to question 1 is ’no’, then the score is <100%. Question 2 has 16 sub-questions, each with an ordinal response. An ordinal, monotonic scaling response (Likert scale) is proposed to rate the respondent’s perspective on evidence quality. Where the response relates to a binary state the extreme response would be expected. The response is weighted from 1 (no bias) to 5 (evidence of bias). The numeric responses should be added to give a summative score that should then be deducted from 100 and expressed as a percentage. The ordinal response informs the extent of impairment in the medical evidence relating to the study. Since accurate and precise measurement of bias may be difficult or impossible, the response is recorded using an ordinal scale. Sample choice bias may be rated based on evidence of exclusion of minority groups (recruitment bias), older groups (age bias) and women (sex bias), or, where there appears to be a clinically meaningful difference in the proportion of the trial population represented by this subgroup compared with the population prevalence.