Urinary Arsenic Metabolism and Birth Outcomes in Tacna, Peru, 2019: A Prospective Cohort Study

Review
I have looked at this paper on arsenic and outcomes for newborns. The paper informs the discussion of where the no-effect level for arsenic sits. In various parts of the world with high arsenic exposure, arsenic has substantial negative impacts on birth outcomes and subsequent health issues. In other places arsenic exposure is lower and outcomes are less severely affected or not affected at all. The paper tries to disentangle some of the complexities involved and although there are no overall effects there are a number of statistically significant findings that are worth noting.

The authors have responded to my earlier review by modifying the text and have also addressed at least some of the points raised by other reviewers. They have also adjusted and explained statistical approaches and extended these as requested - importantly to look at boy-girl differences.

I feel some of the points raised by the group of reviewers taken as a whole would require a quite different paper to be prepared: one that would constitute more of a review of the no-effect literature as well as consideration of the specifics of this study. Such a review would be valuable, especially if it included consideration of the wide range of statistical approaches used in studies of this kind, but to try to do that and deal with the specifics would be very difficult.

I am content that the revised submission can be fully published since it addresses and raises a number of important public health issues even if the study analysis cannot resolve all of these.

Note:
This review refers to round 2 of peer review.

Review
This submission needs some revision before it can be fully-published. Although marked up as needing major revisions I do not think the suggested amendments require too much additional work
by the authors and I think the study could be published after amendments are made.

The study design is clearly focused on exploring the links between birth weight and urinary levels of arsenic. It also discusses some explanation for the “no-effect” result obtained, suggesting that either exposure was not high enough or that there was a protective role within the toxicokinetics - for which there was some pertinent and potentially important data.

Some points for clarification or to help the reader have been included as comments on the submitted manuscripts. These would not be difficult to take into account, adjusting the material accordingly. Marked up documents have been sent to the editorial office by email.

Other points that would improve the paper would be:

It would help the paper to give it a little more context. For example, I wonder if ref 34 provides some of that. Perhaps the results of this study into over 2,000,000 births in Peru in the 21st century ought to be given prominence earlier than it is. It does not consider arsenic at all as far as I am aware and does (as referenced) say birth weights in Tacna are higher – it also suggests why this might be, albeit qualitatively. Making reference to this study earlier might provide a helpful background to the submission as the submission itself raises a potentially important point seemingly not included in ref 34 and also discusses toxicokinetics and demographic differences between the different peoples of Peru – such topics are not often covered in high level studies such as ref 34.

Although the work was designed to address one specific potential relationship and the statistical analysis shows there was “no-effect” or at least no significant relationship as revealed by the statistical analysis, there are a number of significant results in the tables and possibly in the supplementary material. Some of these results are shown in bold in the tables although one is not (see comments on supplementary materials text). It would be helpful to draw these statistically significant results out a little more than has been done in the text – or at least make the link between the significant result and the text clearer. I recognise these significant results, to some degree, sit outside the scope of the main hypothesis but nevertheless they deserve greater prominence especially as they relate to the explanation for “no-effect” put forward.

It should be possible to make more of the point that the paper goes some way to establishing what a “no-effect” level for arsenic exposure might be. There is discussion around this point in the submission – along the lines of making use of references that suggest some effects seen at lower levels of exposure but establishing where no effect levels are seems to be important and again as educational level is known it is potentially possible to link this material to some aspects of ref 34.

It was a surprise to see the data analysed as a group without breaking the analysis into two for boys and for girls. Although weight ranges at birth overlap between the sexes the additional detail, potentially, would improve the paper. If there is an explanation for not doing this please provide the reason. It may be due to sample size and subsequent lack of statistical power, for example.

Lastly, to relate the submission back to ref 34; is it possible to say anything about how the educational status of the mothers might influence the apparent, potentially protective, toxicokinetics, i.e. does toxicokinetics vary with socio-economic status? This might be a step to far for this submission and detract from the existing focus.

Note:
This review refers to round 1 of peer review and may pertain to an earlier version of the document.

Review
The authors explore the association of birth weight to exposure to arsenic, presumably consumed through drinking water. The selected population in Tacna, Peru is exposed to arsenic level of greater than 10 microgram/litre (also ppb) - which is a guideline value described by the World Health Organization (WHO). Based on the study of 158 pregnant women, the authors find no correlation between the arsenic metabolites found in urine samples and the corresponding birthweight.

The authors should consider the following general observations:

1. In the grand scheme, the arsenic exposure through drinking water in Tacna appears to be on the low side - which range from 10 to 25 ppb (based on citation 6, which is the authors' paper published in 2021). In their analysis, the authors compare their findings to those from Bangladesh where the concentrations are often one to two orders of magnitude greater, resulting in marked health impacts including those on pregnancy. The authors should acknowledge this major difference when they undertake such comparisons.

2. The authors do not provide strong evidence for ingestion routes for arsenic. Put differently, they do not present drinking water arsenic concentrations, and there is no attempt at correlating drinking water arsenic concentration to the DMA concentrations, which is a primary arsenic metabolite.

The authors should consider the following revisions:

a. The descriptive paragraph on arsenic in drinking water (lines 10-16) needs to be revised by including a summary of arsenic concentrations found in drinking water, including their speciation (i.e., arsenite vs. arsenate). They should also comment on how these values concentrations compared with those found elsewhere, particularly in Bangladesh/India (partly because they reference Bangladesh in their discussion later in the paper).

b. In the Discussion session, there are several references to arsenic concentration (valued at µg/L) that do not clarify whether they are talking about arsenic concentration in the drinking water or the composite value in urine samples. For example, one can surmise they are referring to arsenic concentration in water on line 131. Whereas the tAs levels listed on line 133 are referring to arsenic in the urine level, one may surmise. This obfuscation, ostensibly not intentional, causes some difficulty for the reader and the chances of miscommunication are great. The authors should revise their presentation of values and clearly distinguish between arsenic concentrations in water and those in urine samples.

c. The statement on line 152, the way it is phrased could be misleading: "Some studies did not find any significant association with adverse birth outcomes, even with exposure levels ≥10 µg/L." 10 µg/L is the lowest end of the adverse value (although no level of arsenic is absolutely safe) identified as a "guideline" by WHO. There should not be an expectation, as implied here, to always observe adverse health impacts any time drinking water concentration goes above 10 µg/L.

d. Some of the explanations provided in the paragraph (line 196-204) seems highly speculative. The authors should consider re-phrasing these statements.

Note:
This review refers to round 1 of peer review and may pertain to an earlier version of the document.