Why an automated tracker finds poor sharing of clinical trial results for an academic sponsor: a bibliometric analysis.