This study will use data from a previously published systematic review about the effectiveness of treatment of chronic non-specific back pain to: describe sample size of published trials in this broad topic area, investigate how trial sample size is related to the trial conduct, size of effect and certainty of evidence, and calculate sample size of new trials that would be required to modify evidence for various treatment comparisons. Our explanatory variable, sample size, will be defined by the smallest arm size in the trial, and descriptive analyses will be conducted by defining a ‘small’ sample size with a threshold of 30 participants. Response variables will include trial conduct (Cochrane Risk of Bias tool); treatment effect (pain intensity and functional limitations); and certainty of evidence (GRADE [Grading of Recommendation, Assessment, Development and Evaluation] recommendations). Our statistical plan will include univariable and multivariable meta-regression analyses to assess the relationship between sample size and trial conduct characteristics, and between sample size and treatment effect, and certainty of evidence. We will also use cumulative meta-analysis to estimate the year in the history of trials published in the field where defined sample sizes stopped contributing information to improve the certainty of evidence. Last, we will use prospective conditional analyses to estimate the “next” trial size and characteristics that would be needed to modify results and interpretation for each comparison. This work will inform future trials in this field regarding what sample sizes can positively contribute to the estimate and certainty of evidence.
