2.6. Risk of Bias in individual studies

Two of the three researchers carried out risk‐of‐bias (ROB) assessments using the Cochrane Collaboration risk‐of‐bias tool v.2.0 that has the following five domains: randomization, deviation from intervention, missing data, measurement of outcome, and selective reporting. ²⁶ Each domain and overall were rated “high risk,” “some concerns,” and “low risk.” For observational studies, we used the Newcastle–Ottawa Scale (NOS) with domains for selection, comparability, and exposure/outcome. ²⁷ Regarding NOS, we assigned (i) age; (ii) diabetes mellitus (DM), chronic kidney disease (CKD), and congestive heart failure (CHF) statuses; and (iii) history of pneumonia as the most important (primary) confounders and (i) sex; (ii) presence of chronic obstructive pulmonary disease (COPD) or asthma; (iii) history of stroke; (iv) ongoing medication (antacid, oral corticosteroid, and other immunosuppressants); (v) smoking status; and (vi) immunization status (against influenza and Streptococcus pneumonia) as secondary confounders. We added one point to the score of comparability of NOS if all primary confounders were adjusted for. An additional point was added if all the secondary confounders were also adjusted for. If the Charlson comorbidity index score was used, we considered age, CHF, CKD, DM, history of stroke, COPD, and asthma to be adjusted. We judged that the history of pneumonia was adjusted if the study looked back over the past year.