On 2017 Nov 16, Erick H Turner commented:

The authors make mention of two approaches for dealing with publication bias. However, both of these can be classified as statistical approaches for use when the only data to which one has access is the published literature. Another approach: simply procure the missing data. Instead of relying on manuscript cohorts, one can look further “upstream” to inception cohorts of clinical trials. Examples of sources include regulatory data (e.g. Drugs@FDA), grants databases, and ClinicalTrials.gov.

On 2017 Nov 16, Erick H Turner commented:

The authors make mention of two approaches for dealing with publication bias. However, both of these can be classified as statistical approaches for use when the only data to which one has access is the published literature. Another approach: simply procure the missing data. Instead of relying on manuscript cohorts, one can look further “upstream” to inception cohorts of clinical trials. Examples of sources include regulatory data (e.g. Drugs@FDA), grants databases, and ClinicalTrials.gov.