PCI Infections Article #90 Decision Round 4

Title: HIV self-testing positivity rate and linkage to confirmatory testing and care: a telephone survey in Côte d’Ivoire, Mali and Senegal 
Arsène Kouassi Kra et al. doi: https://doi.org/10.1101/2023.06.10.23291206 

Dear Authors, 

I commend the work you have put into this important manuscript. I am ready to recommend it, but there is one typo I think the authors will want to fix before I send that out. 

(1) The authors have accidentally deleted a sentence in the discussion (Line 410): 

“For instance, in 2020 an estimated 1.9% of all HIV tests performed were found to be positive in the region (95% credible intervals: 1.3 to 2.7%) [42]. “ 

There are also some additional minor issues, if they wish to take the time to adjust (but not required):

(2) Table 1: “Based on self-interpreted test results” section : High possible rate —> Highest possible rate

(3) There are still issues with references. For instance, Reference 45 is still incomplete. What publication or website is this from? Again a search pulls up a conference poster. Here is an example guideline for citing a poster: https://libguides.ecu.edu/c.php?g=982594&p=7463681 If the authors plan to publish in PCI Journal (which is a great option, since it will be immediately accepted), the references must include the doi where available. 

(4) Re: 1,000 vs 1 000 : Word of advice: Don’t ever “please” reviewers if you can support a decision and are consistent. I live in France and work with WHO AFRO, I am aware of the convention. My issue was that I thought it was inconsistent and the authors had chosen “comma” rather than “space” separators. I was following the tracked-changes version and thought the commas were input rather than deleted (eg, in the first paragraph of the results). Please revert to “spaces” if the authors wish to take the opportunity to do so! Otherwise, it seems consistent now.

Great work. 

Please find my decision and suggestions in the attached PDF. I hope everything is clear and that the authors can easily address these final concerns. 

Sincerely, 
Jessie

Dear authors, 

Thank you for your work on this manuscript, and please find additional comments and concerns raised during this first revision cycle. I hope you will find them clear and useful - if not too terribly detailed. 

Sincerely, 
Jessica Abbate

Decision on PCI Infections Article ID # 90 version 1

HIV self-testing positivity rate and linkage to confirmatory testing and care: a telephone survey in Côte d'Ivoire, Mali and Senegal

Kra Djuhe Arsene Kouassi, Arlette Simo Fotso, Nicolas Rouveau, Mathieu Maheu-Giroux, Marie-Claude Boily, Romain Silhol, Marc d'Elbee, Anthony Vautier, Joseph Larmarange, ATLAS Team

https://doi.org/10.1101/2023.06.10.23291206

I am pleased to be able to provide the authors with feedback on this important study which details how HIV self-testing can help to better reach targeted populations, and which also provides the details of a non-invasive method to help determine the results of that testing. This work is important not just for managing a public health issue, but for tracking the epidemiology of this disease which carries so much stigma and can evade traditional testing methods. While there are some major issues that need to be addressed (see below), I strongly believe these issues are addressable, that the data support the conclusions, and that this article deserves recommendation. I look forward to seeing their revised manuscript. 

Sincerely, 

Jessie Abbate

In addition to the three reviewer comments you will find attached to this decision, I have made some observations myself that need to be addressed. The two main “major” issues I have are as follow:

Major Issue 1: Article structure and scholarship

The article needs to be structured in proper format – with methods in the methods section, results in the results section, and discussion in the discussion. References need to be complete, and the authors should give it more careful review for typos before resubmitting. I would also like to see a bit more scholarship, with previous self-testing methods and studies cited – and those cited explained a bit more. Specific comments to help with this are found below, but also among the other reviewer’s comments.

Major Issue 2: Statistical methods are not clearly stated (i.e., line 201 “explored” how?), and results could use a figure. If I understand correctly, the respondent’s answers to the two phases were linkable despite being anonymous. Thus, it seems Table S1, for instance, should have been done using a repeated measures ANOVA, since results from phase 1 and phase 2 were not independent? I could be wrong about this, but the authors should be more explicit and justify the statistical tests used. It’s not clear what the Chi-sq tests are really testing here. I also agree that clear presentation of analysis related to age groups is needed.

I also agree with reviewers that a figure showing results would be useful. Figures, such as bar charts with bars for standard error of proportion or tukey test results would help the reader understand also what tests were used and how the results grouped (or didn’t).

As Reviewer 1 mentions describing the use of R, the convention is that the functions and packages used for each statistical test (aside from basic summary statistics) should be explicitly written in the methods.

Specific comments:

Lines 30-42: Parroting one reviewer’s comment, I think this part of the methods can be less detailed in the abstract, to be more succinctly presented.

Line 65-66: explain what is 95-95-95 targets.

Seems like the Rouveau etal 2021 paper (ref 23) should be introduced in line 77. Perhaps also ref 41 (Ki-Zerbo et al). 

Please review reference formats, including in-line reference norms, as well as links and accession dates for online materials, such as those published by WHO.

Lines 81-119 should be part of the methods : study design, etc.

Lines 130-135 should be part of the discussion.

Line 178: The data management plan, if approved in French by the appropriate authorities, does not need to be translated, but I agree that it should be stated that it is to be found in French.

Lines 187-190: These two sentences can be combined by simply putting the extra information contained in the second parenthetically into the first. i.e. the low hypothesis considered DK-R as non-reactive (one line), …

Line 203: they type of facility “where” (not “was”) confirmatory testing was performed

Lines 295-297: I think here, the authors should state the fact of suboptimal linkage/care without the negative “however”. Instead, I would write it as a positive “however” on the second sentence:

“Linkage to ….. years of implementation. However, among participants who…”

Line 332: “assumption” => “estimate” (It’s an estimate, based on conservative assumptions)

Lines 331-343: Here, please clarify that the results reported are from the referenced study, and provide more information on that study. How are these lower estimates usually gathered?

Line 349: closer (not close – close would require reporting the confindence intervals for both figures)

Table S2: “I didn’t know WE should get…” not “I didn’t know HE should get…” ?

Table S5: organize the table from shortest time to longest time, rather than alphabetically.

The Discussion needs reorganization. Here are my suggestions, which echo those made by other reviewers:

I would move paragraphs 322-343 up as the first paragraph of the discussion. I would also re-organize, as the second paragraph repeats the assertion while adding more data from other studies. For instance, move lines 337-341 up together with the other conventional estimates listed above. Then finish with lines 324-330 and finally the sentence 341-343.

I would move the paragraph Lines 306-321 to just before the paragraph beginning on line 344 (so, discuss positivity, then interpretation, then confirmatory testing linkage).

Lines 301-305: This should come closer to the end of the discussion, and maybe a bit more detail about the cost should be included. For instance, the fact that phase 2 respondents were not aware of the reward could mean that the cost could potentially be reduced by excluding this reward. I think the authors would do well to say that the alternative methods (explained more: what are the “modeling” methods??) should be compared, including phone surveys without financial incentives.

Lines 294-300 should then be moved to the very end, as a conclusion, to be married with the text currently in the paragraph starting with line 365.