^1. No serious study limitations: all the studies had adequately concealed allocation and blinded both participants and study staff to be considered at low risk of bias. Free of other bias was unclear in Macknin 1998 and Petrus 1998. Petrus 1998 did not adequately describe the sequence generation.

^2. Serious inconsistency: there was high statistical heterogeneity. I² statistic = 93%. The heterogeneity may be due to differences in the nature of the different interventions (zinc gluconate or acetate lozenges, zinc sulphate syrup) and dose range (30 to 160 mg/day) as well as mean duration of symptoms prior to administration of zinc (varying from 24 to 48 hours, as well as the characteristics of the study population (children versus adults). However, subgroup analysis was not possible as there were not enough studies for each variable.

^3. No serious indirectness: studies both from low‐income and high‐income regions have assessed this comparison. Therefore, the result can be confidently generalised to all situations.

^4. No serious imprecision: though the 95% CI around the pooled effect is narrow, the lower limit does not suggest a clinically important reduction in the duration of cold (a decrease in duration of ≤1 day is not shown to be important to patients).

^5. No serious study limitation: all the studies had adequately concealed allocation and blinded both participants and study staff to be considered at low risk of bias. Whether free of other bias was unclear and adequate sequence was not generated in one study (Petrus 1998).

^6. No serious imprecision: though the 95% CI around the pooled effect is narrow, the lower limit does not suggests a clinically important reduction in the severity of symptom score (a change of less than 1 point score is not shown to be important to patients).

^7. Serious inconsistency: there was high statistical heterogeneity. I² statistic = 75%. The heterogeneity may be due to differences in the nature of the different interventions (zinc gluconate or acetate lozenges, zinc sulphate syrup) and dose range (30 to 160 mg/day) as well as mean duration of symptoms prior to administration of zinc (varying from 24 to 48 hours), as well as the characteristics of the study population (children versus adults). However, subgroup analysis was not possible as there were not enough studies for each variable.

^8.Kurugol 2006b is a community‐based intervention including 200 healthy school children and studying the effect of daily administration of 15 mg zinc sulphate syrup over a period of 7 months. Vakili 2009 is also a community‐based intervention including 200 healthy school children and studying the effect of daily administration of 10 mg zinc sulfate tablets over a period of 7 months.

^9. Serious study limitation: though the study by Kurugol 2006b was of high quality, that by Vakili 2009 was of poor methodological quality.

^10. Serious inconsistency: there is substantial heterogeneity between the two trials: I² statistic for heterogeneity = 88%. Both trials showed a benefit with zinc, however the size of this effect was much larger in Vakili 2009. The heterogeneity was due to differences in the trial methodology and the nature of the interventions.

^11. No serious imprecision: the 95% CI around the pooled effect is narrow. Even the lower limit suggests a clinically important reduction in the incidence rate ratio of cold which is shown to be important to patients.

^12. No serious study limitations: allocation concealment was unclear in two studies, i.e. Smith 1989 and Weismann 1990, though both the studies blinded both participants and study staff.

^13. Serious inconsistency: there was high statistical heterogeneity. I² statistic = 75%. The heterogeneity may be due to differences in the nature of the different interventions (zinc gluconate or acetate lozenges) and dose range (30 to 160 mg/day) as well as mean duration of symptoms prior to administration of zinc (varying from 24 to 48 hours, as well as the characteristics of the study population (children versus adults). However, subgroup analysis was not possible as there were not enough studies for each variable.

^14. Serious indirectness: only studies from high‐income regions have assessed this comparison. Therefore, the result can not be generalised to all situations.

^15. No serious imprecision: both limits of the 95% CI suggest a clinically important reduction in proportion of participants given the intervention symptomatic after seven days of treatment.

^16. Serious inconsistency: there is substantial heterogeneity between the two trials: I² statistic test for heterogeneity = 64%. Both trials showed reduced days of school absenteeism with intervention, however, the size of this effect was much larger in Kurugol 2006b. The heterogeneity was due to differences in the trial methodology and the nature of the interventions.

^17. No serious imprecision: though the 95% CI around the pooled effect is narrow, the lower limit does not suggests a clinically important reduction in the duration of school absenteeism (a decrease in duration of ≤1 day is not shown to be important to patients).

^18. No serious inconsistency: there was no statistical heterogeneity. I² statistic = 0%.

^19. No serious imprecision: both limits of the 95% CI suggest a clinically important reduction in the rate of antibiotic use with intervention.

^20. No serious study limitations: all the studies had adequately concealed allocation (except Weismann 1990, in which allocation concealment is unclear) and blinded both participants and study staff to be considered at low risk of bias. Whether free of other bias was unclear in Macknin 1998 and Weismann 1990. Weismann 1990 did not adequately describe the sequence generation.

^21. No serious inconsistency: there was moderate statistical heterogeneity. I² statistic = 51%.

^22. Serious imprecision: the 95% CI around the pooled effect is wide. Though the resulting adverse events from use of zinc is higher, this is not significant.