On 2017 May 30, Rashmi Das commented:

We thank Harri for his PERSONAL (NON PEER REVIEWED) OPINION which is available at above HANDLE ( http://hdl.handle.net/10138/153180) THAT CONTAINS DIRECT COPY AND PASTE OF THREE FIGURES/IMAGES FROM OUR PREVIOUS PUBLICATIONS (JAMA 2014 and Cochrane 2013). We are happy to reply to above comments made by Harri. First regarding the Cochrane review which was withdrawn in 2015, the detailed report is already available at following link (http://onlinelibrary.wiley.com/doi/10.1002/14651858.CD001364.pub5/abstract). This report is the collaborative observation and conclusion of the Cochrane editors (UNLIKE THE HANDLE WHICH CONTAINS MORE OF PERSONAL OPINION WHICH HAS ALREADY BEEN EXAMINED BY THE COCHRANE EDITORS BEFORE REACHING THE CONCLUSION). The same HANDLE WAS SENT TO JAMA EDITORS REGARDING THE JAMA CLINICAL SYNOPSIS (PUBLISHED IN 2014) AND HARRI REQUESTED THE EDITORS TO CARRY OUT THE INVESTIGATION AND VERIFY. THE EDITORS ASKED US FOR REPLY WHICH WE CLARIFIED IN A POINT TO POINT MANNER (BOTH THE COMMENT BY HARRI AND OUR REPLY WAS PUBLISHED, SEE BELOW). HAD THE COMMENT/REPORT BY HARRI WAS ENTIRELY CORRECT, THE JAMA EDITORS COULD HAVE STRAIGHTWAY RETRACTED/WITHDRAWN THE SYNOPSIS WITHOUT GOING FOR PUBLICATION OF THE COMMENT/REPLY (Both are available at following: https://www.ncbi.nlm.nih.gov/pubmed/26284729; https://www.ncbi.nlm.nih.gov/pubmed/26284728). IT HAS TO BE MADE CLEAR THAT THE JAMA SYNOPSIS (DAS 2014) WAS WITHDRAWN AS THE SOURCE DOCUMENT ON WHICH IT WAS BASED (COCHRANE 2013 REVIEW) WAS WITHDRAWN (NOT BASED ON THE REPORT IN THE HANDLE WHICH IS A PERSONAL NON PEER REVIEWED OPINION). The irony is that though HARRI'S COMMENT got published as LETTER TO EDITOR in JAMA after OUR REPLY, still the NON PEER REVIEWED HANDLE THAT CONTAINS DIRECT COPY OF THREE FIGURES/IMAGES FROM OUR PUBLICATION IS GETTING PROPAGATED.

On 2015 Jan 04, Harri Hemila commented:

JAMA Synopsis by Das RR, 2014 is misleading about the effects of zinc on the common cold

Note added in March 2015: After I wrote the comments below, I used time to figure out explanations for the inconsistencies described below. A detailed description of serious problems with the calculations is available as a separate document at: handle.

In 2011, I wrote Feedback about the problems of the Cochrane review (2011) on zinc for the common cold by Singh M, 2011. My Feedback is available at handle, with a summary in Pubmed Commons. Some errors were corrected in the 2013 update by Singh M, 2013, but many remain.

In JAMA, Das RR, 2014 published a Synopsis of the updated Cochrane review (2013). Here I describe the main problems of the JAMA Synopsis.

1) The figures in Table are inconsistent

In the first row of the Table summarizing the zinc lozenge trials, Das RR, 2014 report that the mean duration of cold symptoms was 6.75 (SD 2.36) days in the zinc group and 7.5 (SD 4.03) days in the placebo group, and from these figures Das RR, 2014 calculate the difference as -1.04 (95%CI -2.02, -0.05) days. However, the correct difference between 6.75 and 7.5 is -0.75 instead of -1.04. In addition, the means and SDs give (95%CI -1.10, -0.40) which is much narrower than the 95%CI reported in the Table.

The same problem is seen in the second row of the Table for zinc syrup (2 studies with children). The duration of colds was 5.1 (SD 0.4) days in the zinc group, and 5.9 (SD 0.6) days in the placebo group, the Table giving the difference as -0.63 (95%CI -0.84, -0.43). However, the difference between 5.1 and 5.9 is -0.8 and not -0.63. Furthermore, the means and SDs give (95%CI -0.91, -0.69) which is inconsistent with the Table. Table is also inconsistent with Singh M, 2013, which gives -0.65 (95%CI -0.92, -0.39).

Thus, either the mean durations and their SDs are incorrect, or the estimates and their 95%CIs are incorrect. In any case the figures in the Table are inconsistent.

2) When there is strong heterogeneity, the main focus should be on trying to understand the sources of that heterogeneity, see Thompson SG, 1994

In the Table, Das RR, 2014 report that there is a high level heterogeneity in the effect of zinc lozenges on the duration of colds, with I-square=89%. Previously, Hemilä H, 2011 showed that the heterogeneity over all the zinc lozenge trials can be explained by the dosage of zinc and the zinc salt used in the lozenges. That approach was copied without citation into Fig. 4 in the Cochrane review (2013) by Singh M, 2013, which shows that 5 studies used low doses of zinc, <75 mg/d, and uniformly found no benefit of zinc, whereas 7 studies used high doses of zinc, >75 mg/d, and found a 2.0 day (95%CI 0.8, 3.1; P=0.0006) reduction in the duration of colds. Thus, the dosage of zinc is a source of heterogeneity. However, this plausible explanation for heterogeneity was not made clear to the readers of the JAMA Synopsis.

Furthermore, the third row of the Table shows the comparison “High dose vs low dose zinc lozenges.” No such comparison is published in the Cochrane review (2013). It is not evident where the data for the third row comes from.

Das RR, 2014 state that “Because of… high heterogeneity among the lozenge trials, results should be viewed with caution”, but no justification is given for that statement. When a source of heterogeneity has been identified, heterogeneity per se does not lead to any caution. Low doses of zinc are uniformly ineffective in treating the common cold, but that is no argument against the benefit of high dose zinc lozenges, which seem to reduce the duration of colds by 2 days, see above.

3) There is no basis to speculate that publication bias might explain the reported benefits

Das RR, 2014 claim that ”Because of a significant potential for publication bias … results should be viewed with caution”, but no justification is given for the concern about publication bias in the JAMA synopsis, nor in the Cochrane review (2013).

If we wish to use publication bias as an explanation for the pattern of the published findings, there should be strong correlation between the zinc dosage and the decision to publish the results. When low dose studies uniformly show no evidence of effect (still they were published), and high dose zinc lozenge studies show strong evidence of benefit, a large number of high dose negative studies should remain unpublished, to explain such a pattern. Thus, complex speculation would be needed to explain the pattern of results by publication bias, instead of explaining the pattern as the result of dose dependency.

4) Das RR, 2014 show the effect in days but the percentage effect is more useful for common cold patients

The standard practice of analyzing binary data is to calculate relative effects, which adjusts for baseline variations. In the analysis of common cold duration, relative effects should also be calculated, since these automatically adjust for variation in baseline cold durations.

In Fig. 4 of the Cochrane review (2013), high dose zinc lozenges shorten the duration of colds by 2 days, but we do not know whether the corresponding untreated colds would last for 3 or 14 days. Thus, the practical significance of the reduction by 2 days depends on the baseline duration, but that has varied over the studies. Hemilä H, 2011 calculated that high dose zinc acetate lozenges shortened the durations of colds by 43% (95%CI 35%, 48%) and lozenges made of other salts by 20% (95%CI 12%, 28%). Percentage estimates seem more useful for cold patients since they are not associated with a specific but undefined duration of untreated cold.

5) Das RR, 2014 conclude that “Used prophylactically, oral zinc is associated with a reduced cold incidence in children.” However, the findings on children are from 2 Turkish studies. There is no justification to extrapolate such findings to all children, implying validity for children in developed countries.

6) Das RR, 2014 describe findings of an RCT carried out in Thailand with children, the reporting of which is sloppy, see Pubmed Commons. Even if the findings might be valid for Thai children, they cannot be directly extrapolated to developed countries.

7) Das RR, 2014 claim that “Zinc lozenges were associated with a higher incidence of adverse events compared with placebo.” This conclusion is based on the pooling of studies that used different types of zinc lozenges. However, the adverse effects of lozenges depend on their specific compositions and pooling adverse effects of all zinc lozenge trials is unsound, see comment 9 in Pubmed Commons.

8) Das RR, 2014 conclude their JAMA Synopsis by stating that “further information on the association of zinc dose with toxicity is needed”

For certain patients, zinc has been administered at high doses, 150 mg/d, for therapeutic purposes for months and there are case reports of people taking up to 2000 mg/d of zinc for years. Many of these reports found that long-term high dose zinc decreased copper levels and led to haematological changes, but the changes were reversible with the cessation of zinc intake. See reports of high zinc doses eg by Pories WJ, 1967, Greaves MW, 1970, Simkin PA, 1976, Prasad AS, 1978, Samman S, 1987, Hoffman HN 2nd, 1988, Simon SR, 1988, Forman WB, 1990, Fiske DN, 1994, Irving JA, 2003, Bamford JT, 2012.

Thus, given that months of 150 mg/d zinc regime does not cause permanent harms, it seems evident that treating the common cold for 1 to 2 weeks with 80-90 mg/d of zinc in the form of zinc acetate lozenges does not cause unanticipated harm. Common cold patients should not be scared about the possible toxicity of short-term zinc lozenge treatment.

On 2015 Jan 04, Harri Hemila commented:

JAMA Synopsis by Das RR, 2014 is misleading about the effects of zinc on the common cold

Note added in March 2015: After I wrote the comments below, I used time to figure out explanations for the inconsistencies described below. A detailed description of serious problems with the calculations is available as a separate document at: handle.

In 2011, I wrote Feedback about the problems of the Cochrane review (2011) on zinc for the common cold by Singh M, 2011. My Feedback is available at handle, with a summary in Pubmed Commons. Some errors were corrected in the 2013 update by Singh M, 2013, but many remain.

In JAMA, Das RR, 2014 published a Synopsis of the updated Cochrane review (2013). Here I describe the main problems of the JAMA Synopsis.

1) The figures in Table are inconsistent

In the first row of the Table summarizing the zinc lozenge trials, Das RR, 2014 report that the mean duration of cold symptoms was 6.75 (SD 2.36) days in the zinc group and 7.5 (SD 4.03) days in the placebo group, and from these figures Das RR, 2014 calculate the difference as -1.04 (95%CI -2.02, -0.05) days. However, the correct difference between 6.75 and 7.5 is -0.75 instead of -1.04. In addition, the means and SDs give (95%CI -1.10, -0.40) which is much narrower than the 95%CI reported in the Table.

The same problem is seen in the second row of the Table for zinc syrup (2 studies with children). The duration of colds was 5.1 (SD 0.4) days in the zinc group, and 5.9 (SD 0.6) days in the placebo group, the Table giving the difference as -0.63 (95%CI -0.84, -0.43). However, the difference between 5.1 and 5.9 is -0.8 and not -0.63. Furthermore, the means and SDs give (95%CI -0.91, -0.69) which is inconsistent with the Table. Table is also inconsistent with Singh M, 2013, which gives -0.65 (95%CI -0.92, -0.39).

Thus, either the mean durations and their SDs are incorrect, or the estimates and their 95%CIs are incorrect. In any case the figures in the Table are inconsistent.

2) When there is strong heterogeneity, the main focus should be on trying to understand the sources of that heterogeneity, see Thompson SG, 1994

In the Table, Das RR, 2014 report that there is a high level heterogeneity in the effect of zinc lozenges on the duration of colds, with I-square=89%. Previously, Hemilä H, 2011 showed that the heterogeneity over all the zinc lozenge trials can be explained by the dosage of zinc and the zinc salt used in the lozenges. That approach was copied without citation into Fig. 4 in the Cochrane review (2013) by Singh M, 2013, which shows that 5 studies used low doses of zinc, <75 mg/d, and uniformly found no benefit of zinc, whereas 7 studies used high doses of zinc, >75 mg/d, and found a 2.0 day (95%CI 0.8, 3.1; P=0.0006) reduction in the duration of colds. Thus, the dosage of zinc is a source of heterogeneity. However, this plausible explanation for heterogeneity was not made clear to the readers of the JAMA Synopsis.

Furthermore, the third row of the Table shows the comparison “High dose vs low dose zinc lozenges.” No such comparison is published in the Cochrane review (2013). It is not evident where the data for the third row comes from.

Das RR, 2014 state that “Because of… high heterogeneity among the lozenge trials, results should be viewed with caution”, but no justification is given for that statement. When a source of heterogeneity has been identified, heterogeneity per se does not lead to any caution. Low doses of zinc are uniformly ineffective in treating the common cold, but that is no argument against the benefit of high dose zinc lozenges, which seem to reduce the duration of colds by 2 days, see above.

3) There is no basis to speculate that publication bias might explain the reported benefits

Das RR, 2014 claim that ”Because of a significant potential for publication bias … results should be viewed with caution”, but no justification is given for the concern about publication bias in the JAMA synopsis, nor in the Cochrane review (2013).

If we wish to use publication bias as an explanation for the pattern of the published findings, there should be strong correlation between the zinc dosage and the decision to publish the results. When low dose studies uniformly show no evidence of effect (still they were published), and high dose zinc lozenge studies show strong evidence of benefit, a large number of high dose negative studies should remain unpublished, to explain such a pattern. Thus, complex speculation would be needed to explain the pattern of results by publication bias, instead of explaining the pattern as the result of dose dependency.

4) Das RR, 2014 show the effect in days but the percentage effect is more useful for common cold patients

The standard practice of analyzing binary data is to calculate relative effects, which adjusts for baseline variations. In the analysis of common cold duration, relative effects should also be calculated, since these automatically adjust for variation in baseline cold durations.

In Fig. 4 of the Cochrane review (2013), high dose zinc lozenges shorten the duration of colds by 2 days, but we do not know whether the corresponding untreated colds would last for 3 or 14 days. Thus, the practical significance of the reduction by 2 days depends on the baseline duration, but that has varied over the studies. Hemilä H, 2011 calculated that high dose zinc acetate lozenges shortened the durations of colds by 43% (95%CI 35%, 48%) and lozenges made of other salts by 20% (95%CI 12%, 28%). Percentage estimates seem more useful for cold patients since they are not associated with a specific but undefined duration of untreated cold.

5) Das RR, 2014 conclude that “Used prophylactically, oral zinc is associated with a reduced cold incidence in children.” However, the findings on children are from 2 Turkish studies. There is no justification to extrapolate such findings to all children, implying validity for children in developed countries.

6) Das RR, 2014 describe findings of an RCT carried out in Thailand with children, the reporting of which is sloppy, see Pubmed Commons. Even if the findings might be valid for Thai children, they cannot be directly extrapolated to developed countries.

7) Das RR, 2014 claim that “Zinc lozenges were associated with a higher incidence of adverse events compared with placebo.” This conclusion is based on the pooling of studies that used different types of zinc lozenges. However, the adverse effects of lozenges depend on their specific compositions and pooling adverse effects of all zinc lozenge trials is unsound, see comment 9 in Pubmed Commons.

8) Das RR, 2014 conclude their JAMA Synopsis by stating that “further information on the association of zinc dose with toxicity is needed”

For certain patients, zinc has been administered at high doses, 150 mg/d, for therapeutic purposes for months and there are case reports of people taking up to 2000 mg/d of zinc for years. Many of these reports found that long-term high dose zinc decreased copper levels and led to haematological changes, but the changes were reversible with the cessation of zinc intake. See reports of high zinc doses eg by Pories WJ, 1967, Greaves MW, 1970, Simkin PA, 1976, Prasad AS, 1978, Samman S, 1987, Hoffman HN 2nd, 1988, Simon SR, 1988, Forman WB, 1990, Fiske DN, 1994, Irving JA, 2003, Bamford JT, 2012.

Thus, given that months of 150 mg/d zinc regime does not cause permanent harms, it seems evident that treating the common cold for 1 to 2 weeks with 80-90 mg/d of zinc in the form of zinc acetate lozenges does not cause unanticipated harm. Common cold patients should not be scared about the possible toxicity of short-term zinc lozenge treatment.