Part 1: methodology

Since Eby's trial in 1984 (Eby 1984), many trials have investigated whether zinc is efficacious in the treatment or prevention of the common cold. Among these 18 trials were included in this 2013 review. We rated the methodological quality of the included trials as good, with two trials excluded because of poor quality. Eby's trial realised a number of limitations which raised concerns regarding the validity of the results. Treatment blinding in the trial has been questioned as zinc lozenges were found to be unpalatable, distorted the taste of participants and caused a higher incidence of side effects. In addition, investigators relied solely on the subjective assessment of cold symptoms; laboratory confirmation of viral infection was not conducted and analyses were only conducted on a subgroup of those originally enrolled in the trial. Eby's trial was nevertheless instructive and highlighted a number of methodological issues.

Like Eby's trial, most trials have relied on community‐acquired infections. However, five trials recruited participants from volunteers experimentally inoculated with human rhinovirus (Al‐Nakib 1987; Farr 1987a; Turner 2000a; Turner 2000b; Turner 2000c). While high rates of infection with human rhinovirus were attained in the later trials and most participants experienced cold symptoms, in trials relying on community‐acquired infection, the infecting agent and the infection rates were generally not determined.

In trials relying on community‐acquired infection, investigators relied on trial participants or family members to assess the incidence and severity of cold symptoms. Though in most of the trials information was generally provided on how compliance with the recording of symptoms was assessed, objective periodic assessments of the clinical severity of respiratory symptoms were not conducted. In the trials conducted by Al‐Nakib 1987, Farr 1987a, Macknin 1998, Kurugol 2006a, Kurugol 2006b, Kurugol 2007 and Turner 2000a, Turner 2000b and Turner 2000c, symptoms were assessed by trial personnel thus providing some assurance as to the validity of clinical severity scores and estimates based on such scores. Assessment of response to treatment also depended on objective measurements such as nasal mucus weight or tissue counts, which was measured in one study (Al‐Nakib 1987) and the authors found that zinc gluconate reduced both of these parameters. However, as in most studies children were involved, this was not practical.

Research by Farr 1987b suggested that in most trials the size of the placebo‐blinding study used to determine whether zinc and placebo lozenges were indistinguishable was not sufficiently large to detect a significant difference. In their efforts to find a suitable matching placebo lozenge, Farr showed that a false negative result may occur if a small subject population (i.e. fewer than 20) is used. Given that placebo‐blinding studies were only conducted in six of the 18 trials, and with the exception of three trials (Farr 1987a; Prasad 2000; Prasad 2008) the size of the placebo‐matching studies in two of the remaining three trials (no information was provided on the size of placebo study conducted by Weismann 1990) ranged from eight to 20, the adequacy of blinding in most trials is questioned. Zinc‐treated participants also experienced higher incidences of side effects, complaints or both, and in seven trials zinc‐treated participants complained of altered, bad or unpalatable taste, which suggests that zinc lozenges were distinct from placebo lozenges and, in this respect, blinding may have been compromised. However, the increased incidence of bad taste and nausea found by Mossad 1996; constipation and mouth dryness found by Prasad 2000; and bad taste, nausea, mouth, tongue or throat discomfort and diarrhoea found by Macknin 1998, may have been related to the use of different ligands (gluconate, acetate) rather than to zinc itself.

Much of the controversy surrounding the use of zinc lozenges in the treatment of the common cold has concerned whether formulations used in trials showing no benefit failed to release sufficient zinc ions to be effective. It has been hypothesised that there is a direct correlation between reductions in the duration of common cold symptoms and the daily dosage of all positively charged zinc species released from lozenges at physiologic pH (Eby 1995). The reanalysis of 10 double‐blind, placebo‐controlled zinc trials by solution chemistry methods showed a significant correlation between total daily dosages of positively charged zinc species and a reduction in the mean duration of common colds (Eby 2004). Zinc gluconate and zinc acetate have very low chemical stability and mainly release positively charged zinc ions in aqueous solutions at physiologic pH, but stronger complexes do not (Eby 2004). Adding a strong zinc‐binding ligand, such as glycine, citric acid, tartaric acid, mannitol and sorbitol, to a solution containing a zinc complex that is weakly bonded results in the sequestration of zinc to the stronger ligand, reducing or eliminating the benefits of zinc lozenges (Eby 2004). The extent to which the zinc ion was released from formulations reporting no benefit is not known. However, experimental evidence suggests that in saliva the ionisation of zinc to free zinc for some formulations may have been completely (Zarembo 1992) or partially (Farr 1988) suppressed. A formulation developed by Godfrey (Godfrey 1992) that incorporates glycine has been shown to release more than 50% of the zinc from zinc gluconate as the zinc ion. Results from trials conducted by Godfrey (Godfrey 1992) and Mossad (Mossad 1996) suggest this formulation reduced the duration and severity of respiratory symptoms; whereas the trials conducted by Macknin (Macknin 1998) and Turner (Turner 2000a; Turner 2000b; Turner 2000c) suggest no effect of this formulation. The placebo‐matching was inadequate; consequently the adequacy of blinding in all these trials is questioned. Hatch et al (Hatch 1987) reported that zinc acetate releases essentially 100% of its zinc as Zn²⁺ ions at a physiological pH. Thus, use of zinc lozenges may be advantageous. Results from trials conducted by Petrus (Petrus 1998) and Prasad (Prasad 2000; Prasad 2008) suggest this formulation reduced the duration and severity of respiratory symptoms. These studies used compressed lozenges designed by George Eby that were identical in composition. In addition to zinc acetate, they contained directly compressible (agglomerated) dextrose as the tablet base, glycerol monostearate (2.5% of tablet weight) as the tablet lubricant, stevia for added sweetness, and peppermint oil for flavour, with the composition compressed to near maximal hardness for slowest dissolution. These ingredients were specifically chosen because they do not react with ionic zinc. The trials conducted by Douglas (Douglas 1987) and Turner (Turner 2000a; Turner 2000b; Turner 2000c) using zinc acetate suggest no effect of this formulation. Among these, the placebo‐matching and blinding were stated to be adequate in all except the trial by Turner. Three trials (Kurugol 2006a; Kurugol 2006b; Kurugol 2007) used syrup preparation of zinc (zinc sulphate) and found reduced duration and severity of respiratory symptoms without any increase in adverse effects in the zinc group. Placebo‐matching and adequacy of blinding was not stated in these trials. Another trial (Vakili 2009) used tablet preparation (zinc sulphate) and found decreased incidence, fewer school absences, less antibiotic administration and no adverse effects in zinc‐supplemented children. Again placebo‐matching and adequacy of blinding was not stated in this trial.

The toxicology of zinc has been well characterised. The potential for elevated blood levels of zinc to disrupt copper metabolism and other nutrients preclude its long‐term use in the treatment of the common cold (Pfeiffer 1980). Doses higher than 150 mg/day have also been associated with adverse effects (Chandra 1984). In addition, the higher incidence of side effects in zinc‐treated participants will most likely limit the usefulness of zinc in the treatment of common cold symptoms. In 14 trials (Douglas 1987; Kurugol 2006a; Kurugol 2006b; Kurugol 2007; Macknin 1998; Mossad 1996; Prasad 2000; Prasad 2008; Smith 1989; Turner 2000a; Turner 2000b; Turner 2000c; Vakili 2009; Weismann 1990) included for reporting of any or individual adverse events, the overall adverse events were higher in the intervention than in the control group, except in one trial (Vakili 2009).

Viral studies were performed in nine trials (Al‐Nakib 1987; Douglas 1987; Farr 1987a; Kurugol 2006a; Kurugol 2006b; Kurugol 2007; Turner 2000a; Turner 2000b; Turner 2000c). While in‐vitro studies suggest zinc inhibits viral replication and the concentration of zinc in saliva should be sufficient to induce such an effect, six trials (Al‐Nakib 1987; Douglas 1987; Farr 1987a; Turner 2000a; Turner 2000b; Turner 2000c) found no effect of zinc on incidence or shedding of rhinovirus by study participants. In relation to this, trials by Farr and Douglas found no effect of zinc in the treatment of cold, while Al‐Nakib found reduction in the clinical symptom score. From the trial by Al‐Nakib, it might be suggested that medication may have had an effect on signs and symptoms of the colds rather than on virus replication. If this is the case, it would be interesting to know whether zinc would also have the same effect on corona virus colds or, indeed, on colds caused by other respiratory viruses. This may be the future area of research in zinc and common cold trials. Trials by Kurugol (Kurugol 2006a; Kurugol 2006b; Kurugol 2007) did not study the effect of zinc on rhinovirus cold; rather they excluded colds due to influenza viruses from analysis and found that zinc is effective in the treatment of the common cold.

Part 2: results

Although most investigators required participants to record the clinical severity of symptoms each day and used similar scales against which to rate symptom severity (symptoms were rated as none, mild, moderate or severe), there was little commonality in summary estimates used by investigators to describe the duration, incidence and severity of respiratory symptoms. Consequently for some outcomes it was not possible to pool the results from all the trials that reported the same outcome. In addition, there was insufficient detail provided in most published papers to determine whether trials used similar criteria for rating the severity of each symptom and therefore it was not possible to standardise clinical severity scores across all trials.

Among the trials reporting the duration of cold symptoms, results could be pooled from 14 trials. Intake of zinc lozenges or syrup was associated with a significant reduction in the duration of common cold, but there was marked heterogeneity among the included trials. Among these, two trials (Kurugol 2006a; Kurugol 2007) used syrup preparation at a daily dose of 30 mg, whereas other trials used lozenge preparation at a daily dose varying from 30 mg to 190 mg. In a subgroup analysis, the results were significant for the dose of zinc lozenges ≥ 75 mg/day, any formulation of zinc (lozenges or syrup) and any age group (children or adult). Results were not significant for any formulations of zinc lozenges (gluconate or acetate). However, the heterogeneity persisted for all these subgroups except syrup formulation of zinc. Among the two trials using syrup preparation, one trial (Kurugol 2007) found no significant effect, and the other found significant reduction of the duration of common cold. There has been speculation regarding how zinc cures common cold. The most biologically plausible hypothesis, as well as experimental, was by Eby (Eby 2010; Eby 2012). As described by him, the mouth–nose act as a biologically closed electric circuit (BCEC). It moves electrons from the nose into the mouth and, in response to the electron flow, it moves positively charged metal ions, such as ionic zinc, from the mouth into the nose. This BCEC does not transport neutrally or negatively charged zinc into the nose; so the lozenges made with non‐positively charged zinc are supposed not to work in common cold therapy. Considering these properties, the observation that zinc lozenges releasing positively charged ionic zinc shorten colds in a dose–response manner can be seen in a more plausible light. This is also supported by finding of usefulness of ≥ 75 mg/day of zinc lozenges in reducing the duration of common cold in the present review. The BCEC also explains why patients with colds and a history of allergy responded so much faster than patients with colds and no history of allergy, as shown by Petrus (Petrus 1998). Orally ingested forms such as tablets, liquids and syrups do not have these properties and therefore, according to Eby's research, are not effective therapy for the common cold (Eby 2012).

Among 13 trials measuring the severity score of cold symptoms, results from five trials (Kurugol 2006a; Kurugol 2007; Petrus 1998; Prasad 2000; Prasad 2008) could be pooled. There was a significant difference between the intervention and control group for reduction in the severity of cold symptoms. The trials using syrup preparation used a daily dose of 30 mg, whereas the trials using lozenge preparation used daily doses varying from 30 mg to 276 mg. Al‐Nakib (Al‐Nakib 1987) and Douglas (Douglas 1987) provided estimates of mean clinical scores. However, estimates were not directly comparable. In the trial by Al‐Nakib, the zinc group had a significantly lower mean daily clinical score than the placebo group, with the difference in scores attaining statistical significance on days four and five. However, in the trial conducted by Douglas, there were no statistically significant differences between zinc and placebo groups with respect to mean nasal, throat and cough scores. Results of the trial conducted by Godfrey (Godfrey 1992) suggested treatment with zinc reduced the frequency and severity of cold symptoms, which was noticeable by day five and significant by day seven. In the trial conducted by Turner (Turner 2000a; Turner 2000b; Turner 2000c) none of the zinc preparations (neither gluconate nor acetate) had any significant effect on the severity of common cold symptoms in the first three days of treatment, which might be related to the lower dose of zinc used in this trial. Among four trials (Douglas 1987; Kurugol 2006a; Kurugol 2007; Petrus 1998) measuring individual symptom scores there was a significant reduction in the cough score, with nasal and throat score being variably affected. In the trial conducted by Smith 1989, the zinc group had lower symptom severity scores on days four to seven of treatment which was statistically significant (P = 0.02); but in the trial conducted by Weismann 1990, no statistically significant differences between the two groups were found by day six of treatment (P = 0.14).

Among the two preventive trials measuring the incidence of the common cold (Kurugol 2006b; Vakili 2009), the incidence rate ratio (IRR) of developing a cold in participants who received the intervention was lower than in the placebo group. There was marked heterogeneity, therefore we used a random‐effects model for analysing this outcome. The second trial, though a randomised controlled trial (RCT), was not of good methodological quality, but this was included in the analysis as it included a large number of participants. Even after excluding this trial from analysis, the result still favoured zinc supplementation. The first trial used zinc sulfate syrup at a daily dose of 15 mg for seven months, whereas the second trial used zinc sulfate tablet at a daily dose of 10 mg for five months.

The proportion of participants who were asymptomatic after three, five and seven days of treatment was reported in the trials conducted by Mossad 1996, Weismann 1990 and Smith 1989, and the proportion asymptomatic after seven days of treatment was reported in all but the trials conducted by Farr 1987a and Al‐Nakib 1987. Analyses were conducted on an intention‐to‐treat (ITT) basis. In the trial conducted by Mossad (Mossad 1996) participants were less likely to have cold symptoms after three and five days of treatment in the zinc‐treated group. The odds ratios (ORs) for days three and five were 0.37 (95% confidence interval (CI) 0.14 to 0.92) and 0.35 (95% CI 0.16 to 0.76), respectively. In the trials conducted by Weismann 1990 and Smith 1989, the proportion of participants asymptomatic after three and five days in the zinc and placebo groups was similar. The test for heterogeneity attained statistical significance for day five, but not day three and consequently a combined OR for day five is not appropriate. The combined OR for day three was not significant 0.97 (95% CI 0.62 to 1.5). In three trials (Godfrey 1992; Mossad 1996; Weismann 1990), fewer participants in the zinc group had cold symptoms after seven days. The pooled result (OR 0.53, 95% CI 0.38 to 0.75) obtained from five trials (Douglas 1987; Godfrey 1992; Mossad 1996; Smith 1989; Weismann 1990) indicated fewer participants in the zinc group had cold symptoms after seven days of treatment. However, the test for heterogeneity was statistically significant.

In six trials (Godfrey 1992; Kurugol 2006a; Mossad 1996; Petrus 1998; Prasad 2000; Prasad 2008) with similar study designs, methodologies and efficacy assessments, zinc was found to be effective in reducing the duration and severity of common cold symptoms in healthy children and adults, when it was administered within 24 hours of the onset of symptoms. In another trial (Kurugol 2007) with similar study design, methodology and efficacy assessments, zinc was found to be effective in reducing the severity of common cold symptoms in healthy children (without any change in duration), when it was administered within 24 to 48 hours of the onset of symptoms. In the trial by Godfrey 1992, the authors found a significant decrease in the duration and severity of symptoms when treatment was administered within 24 hours, compared to treatment administration within 48 hours. So, if treatment with zinc is to be used for common cold, it should ideally commence within 24 hours of onset of symptoms.

There are a number of potential sources of heterogeneity in results obtained from trials included in this review. Most trials relied on community‐acquired infections in which the infecting agent was not identified and as such different agents may have been involved which may have differed in their sensitivity to zinc. The amount of zinc taken each day by participants varied largely across the trials, and given that some formulations released less zinc ion than others the effective dose of zinc across trials was variable. Blinding of treatment may not have been adequately controlled in some trials, thereby increasing the potential for performance and detection bias to occur. The time from onset of cold symptoms to commencement of treatment ranged from one to three days. Given the beneficial effects noted in trials commencing treatment with zinc within 24 hours, the results from all the trials may not be comparable. Last but not the least is the fact that the lifestyle of the study population in all the trials was different and the results might have been affected to some degree by this factor.

Summary of main results

Studies reporting duration and severity of cold symptoms suggest that the intake of zinc is associated with a significant reduction in the overall duration and severity of common cold symptoms. A higher proportion of participants became asymptomatic by day seven of treatment with zinc. Duration of individual cold symptoms was also significantly reduced in the zinc group, though the individual symptom severity scores were not significantly affected by the intake of zinc. Zinc supplementation led to reduction in the incidence of common cold, decreased school absence and decreased the risk of antibiotic use when used for at least five months. The incidence of adverse events was significantly higher in the zinc group with the syrup preparation being better tolerated than lozenges.

Overall completeness and applicability of evidence

The trials included in the analysis involved healthy children and adults of all ages (except infants) and both sexes. All except four trials (Kurugol 2006a; Kurugol 2006b; Kurugol 2007; Vakili 2009) were conducted in upper‐middle and high‐income countries, where zinc deficiency is uncommon (including in young children). Out of four trials (Kurugol 2006a; Kurugol 2006b; Kurugol 2007; Vakili 2009), three were conducted in Turkey and one in Iran, which are low‐income countries. These later trials found beneficial effects of zinc when used either in syrup or tablet (not lozenges) form. So, any beneficial effect of zinc lozenges might not be directly extrapolated to children and adults of low‐income countries. In all the included trials, the main weakness of the data is that most trials, when presenting data, did not differentiate between cold due to rhinovirus and other viruses (as viral studies were not conducted in most of these trials). So it is unclear whether zinc helps those with rhinoviral cold or even cold due to other viruses. However, as rhinovirus is the most common aetiological agent of the common cold all over the world (in both low‐income and high‐income countries), it may be predicted that zinc lozenges prepared in the above mentioned formulations might also help people living in low‐income countries.

Quality of the evidence

The trial evidence included is generally of good quality, with a low risk of bias. All the studies were blinded, but placebo‐blinding was adequately described in only six trials. In 12 trials, zinc‐treated participants complained of altered, bad or unpalatable taste which suggests that zinc lozenges were distinct from placebo lozenges and, in this respect, blinding may have been compromised. Allocation concealment was adequate in nine studies and unclear in nine studies. Thirteen trials reported a low rate of loss to follow‐up, whereas three trials did not mention any loss to follow‐up. This suggests that the studies were of good quality. For all the outcomes, there was more than one study reporting the individual outcome. The majority were carefully conducted community trials, with active mechanisms to promote adherence to the intervention and both active and passive case finding. Last and most importantly, as there was a high degree of heterogeneity among the trials, the result should be interpreted with caution.

Potential biases in the review process

All the included trials, as expected, measured the effect of zinc on the common cold. There was therefore the potential to miss trials which may have measured common cold as upper respiratory tract infection in secondary outcomes which were less publicised or less well indexed within the electronic databases. We tried to avoid this by conducting a wide search and assessing the relevance of each paper identified in that search carefully. We extracted data from survival curves or t/P values given in studies that did not provide original data, which might have overestimated or underestimated the effect size. Lastly, publication bias cannot be ruled out as shown by funnel plot asymmetry.

Agreements and disagreements with other studies or reviews

The important changes in this updated review in comparison to the previous version (Marshall 1999) include the following.

1. Intake of zinc is associated with a significant reduction in the duration and severity of common cold symptoms.

2. Duration of individual cold symptoms was also significantly reduced in the zinc group.

3. The syrup and tablet preparation of zinc is better tolerated than lozenges.

4. Zinc supplementation reduces incidence, school absence and prescription of antibiotics in children with the common cold.

In the review conducted by Marshall (Marshall 1999) the included studies had missing information, due to which it was not possible to pool the results across the studies. For example, although most investigators required participants to record the clinical severity of symptoms each day and used similar scales against which to rate symptom severity (symptoms were rated as none, mild, moderate or severe), there was little commonality in the summary estimates used by investigators to describe the duration, incidence and severity of respiratory symptoms. It was therefore only possible to determine the proportion of participants who were asymptomatic after three and five days of treatment for three trials and after seven days of treatment for five trials. Except for one study (Godfrey 1992), duration of symptoms was not reported in the rest of the studies. Therefore, again it was not possible to pool the results for this outcome.

In the revisit of their previous meta‐analysis, Jackson (Jackson 2000) found statistically significant heterogeneity between the zinc trials. They calculated a pooled estimate of the zinc gluconate lozenges in colds using the random‐effects model of DerSimonian and Laird from eight trials. The summary OR for the presence of "any cold symptoms" at seven days was 0.52 (95% CI 0.25 to 1.2). They concluded that despite numerous randomised trials, the evidence for the effectiveness of zinc lozenges in reducing the duration of common colds is still lacking. Although, according to them, some of the negative results might be explained by low zinc ion availability, they did not examine the issue. Moreover, they have not looked for the effect of any other formulations, as well as high or low doses of zinc.

A review published in 2004 by Hulisz 2004, which was an overview of published articles through MEDLINE (1980 to 2003), concluded that zinc effectively reduces the duration and severity of common cold symptoms when administered within 24 hours of the onset of symptoms. The author also had some concerns regarding the clinical tests of zinc for the treatment of common colds being inconsistent, primarily because of study design, blinding and lozenge contents; early formulations of lozenges being unpalatable with a higher incidence of side effects.

In a systematic review of trials published between 1966 and 2006, Caruso 2007 used 11 features of experimental design affecting signal quality, chance, bias and blinding to evaluate 14 placebo‐controlled trials. They gave one point for each of the 11 quality items if it was satisfied, after which only four trials obtained the maximum of 11 points. Based on this, they suggested that the positive results in the zinc trials might be explained by methodological flaws. However, the use of scales for assessing quality or risk of bias is strongly discouraged in Cochrane systematic reviews. Moreover, such an approach, though simple, is not supported by evidence (Higgins 2011).

In another systematic review by Hemilä (Hemilä 2011), the author included 13 placebo‐controlled comparisons. Pooling of trials using a total daily zinc dose of less than 75 mg found no effect, but those using zinc acetate in daily doses of over 75 mg found a 42% reduction, and those using zinc salts other than acetate in daily doses of over 75 mg found a 20% reduction in the duration of colds. The author concluded that there is strong evidence that the zinc lozenge effect on common cold duration is heterogeneous so that benefit is observed with high doses of zinc but not with low doses. However, there were no descriptions of other formulations of zinc (syrup or tablet).

In the latest systematic review (Science 2012), the authors included 17 trials. Compared with patients given placebo, those receiving zinc had a shorter duration of cold symptoms (mean difference (MD) ‐1.65 days, 95% CI ‐2.50 to ‐0.81). However, heterogeneity was high. Zinc shortened the duration of cold symptoms in adults but no significant effect was seen among children. In contrast to this, we found zinc to be effective both in children (< 16 years age) and adults. The reduction in the duration of cold symptoms was greater with high doses of ionic zinc (MD ‐2.75, 95% CI ‐3.89 to ‐1.60) than with lower doses (MD ‐0.84, 95% CI ‐1.50 to ‐0.18). However, we found that only zinc lozenges containing high dose (≥ 75 mg/day) reduced the cold duration significantly.