Model Development

The scoring method converts each pooled relative risk to beta using the natural logarithm, multiplies by 10, and rounds to an integer. Predictors will be prioritised by frequency in independent studies, pooled relative risk magnitude, and guideline-endorsed clinical accessibility. Predictors associated with response receive positive scores when relative risk is above 1, while risk factors below 1 receive negative scores. The protocol will develop nested models A through K, adding predictors sequentially from simpler to more complex models. Candidate predictors will be drawn from meta-analysis findings and supported by subgroup and sensitivity analyses.