In this article by Robert Yeh and colleagues (BMJ 2018;363:k5094, doi:10.1136/bmj.k5094), the final sentence of the Methods section should have read: “A P value less than 0.05 was considered statistically significant.” Both print and online versions will be amended.

