NOTICE: This document contains comments from the reviewers and editors generated during peer review of the initial manuscript submission and sent to the author via email.
Questions about these materials may be directed to the Obstetrics & Gynecology editorial office:
[email protected].
Date: Nov 01, 2019 To: "Nansi Boghossian"
From: "The Green Journal" [email protected] Subject: Your Submission ONG-19-1890
RE: Manuscript Number ONG-19-1890
Regional and Racial/Ethnic Differences in Perinatal Interventions among Periviable Gestations Dear Dr. Boghossian:
Your manuscript has been reviewed by the Editorial Board and by special expert referees. Although it is judged not acceptable for publication in Obstetrics & Gynecology in its present form, we would be willing to give further consideration to a revised version.
If you wish to consider revising your manuscript, you will first need to study carefully the enclosed reports submitted by the referees and editors. Each point raised requires a response, by either revising your manuscript or making a clear and convincing argument as to why no revision is needed. To facilitate our review, we prefer that the cover letter include the comments made by the reviewers and the editor followed by your response. The revised manuscript should indicate the position of all changes made. We suggest that you use the "track changes" feature in your word processing software to do so (rather than strikethrough or underline formatting).
Your paper will be maintained in active status for 14 days from the date of this letter. If we have not heard from you by Nov 15, 2019, we will assume you wish to withdraw the manuscript from further consideration.
REVIEWER COMMENTS:
Reviewer #1:
OVERALL
The study was designed to evaluate the regional and racial differences in perinatal interventions among periviable gestations. This has been a subject that is important to perinatal care and needs to be further investigated.
The reviewer has three major concerns regarding this draft. First, the rationales of conducting this study need to be emphasized. Second, information about the data source and statistical analyses was not provided enough. Third, the tables in the Result section are quite confusing and needs to be re-structured. The comments/suggestions are listed below:
ABSTRACT
Please provide p-values in the Result section INTRODUCTION
Please provide the rationales of choosing region as the key testing variable in this current study. Readers need to know more about this.
The reviewer assumed that the ACOG treatment guideline is national. The implication of this guideline is national. Why could physicians in each U.S. region have so different perspective and practice habits that make the authors think the region factor is so important to be investigated. Please explain/clarify.
For example, the authors chose region but not each state as the key testing variable.
The same concern is similar to the other key testing variable, race/ethnicity. Please provide additional background information.
"However, these studies … the region of practice." This statement is not clear. Is there any reference that can support this statement? Reference 6, 7 seems to only talk about race but not region.
METHOD
Data source
The description of the data source was not clear enough. For example, more information about the NICU and Vermont Oxford Network needs to be provided.
Is the data source from the administrative claims data, register data, or EMR data? Please clarify.
Statistical analysis
Has the basic assumption of the Poisson regression been tested? Did the authors consider to use negative binomial regression model? Please clarify.
RESULTS
The reviewer assumed that region is the key testing variable. However, the way the authors made the tables (eTable 1-5) makes the reviewer think the region is the dependent variable in this study. This is very confusing. Please clarify.
DISCUSSION
"While country-to-country variation … at 23 week' gestation." These statement can become a part of the Introduction section.
Reviewer #2:
Precis - Regional and racial/ ethnic differences in provision of antenatal steroids (ANS), c/s, and postnatal life support (PLS) at 22 -23 weeks
Abstract - Objective - to examine 1) regional differences in perinatal interventions that reflect active treatment (steroids, c/s, PLS) among periviable gestations; 2) racial/ethnic differences of these 3 perinatal interventions by region
Methods - 776 hospitals between 2006-2017 - gestational age 22-25 outcomes - antenatal steroids, c/s, life support
Results - major regional variation in the use of antenatal steroids, c/s PNS between 22-23 weeks midwest and south with higher rates than northeast and west
Conclusions - regional variation in perinatal intervention when managing infants 22-23 weeks - future studies needed to address causes of differences
Intro - periviable birth guidelines - consider steroids /resuscitation at 23 wk, recommend at 24 and 25 weeks; avoid c/s at 22 wk, consider at 23 and 24 weeks, recommend at 25 weeks
previous studies showed minorities had more interventions but there was no accounting for differences in regional practices - intent was to study regional differences and racial/ ethnic differences
Methods - Study population I- retrospective study between 22-25 weeks, 776 NICUs between 2006-2017 Study variables - self reported race/ethnicity - birth certificate and medical record review
examined outcomes - antenatal steroids, c/s, PLS and examined survival of lovelorn up to 1 year Statistical analysis -
1) rate of ANS, c/s, PLS by region 2) survival rates by region
3) ratio of ANS, c/s, PLS by region/ race/ethnicity 4) RR for ANS, c/s, PLS, by race ethnicity
22 and 23 weeks results - primary outcome and 24-25 weeks - secondary outcome Results. study sample - 99,143 infants between 22-25 weeks
Regional variation - ANS, c/s, PLS
across hospitals by region at 22 and 23 weeks - midwest and south have a higher intervention of all 3 than Northeast and West - less variation at 24 and 25 weeks
Regional variation - survival - increase PLS reflected in survival at d1 of life Racial / ethnic differences - ANS, c/s, PLS - 22 and 23 wks
- 22 wk - no difference between racial groups - 23 wk - difference in racial variations by region
- PLS - difference between race and region - difference decreases as approach 24-25 weeks adjusted risk - ANS, c/s, PLS - ANS, PLS and c/s varied by race/ region
Discussion - >99,000 infants between 22-25 wks - major regional variation exists between ANS, c/s, PLS midwest and South had higher rates for all 3 interventions
racial/ regional differences were less apparent at 24-25 wks wide regional variability is due to a lack of robust guidelines
study of SMFM MDs in midwest and south - more willing to intervene in periviable gestation no data on cause of regional variation and decision making process
major regional variation is present between 22-23 weeks
Comments. It is interesting to see that the differences are not necessarily by region but actually by race - I would be interested in knowing the following:
1) if you combine all groups from all regions to eliminate regional differences - is there a difference based on race
2) please clarify and describe racial variation as a percentage of population - ie, if blacks comprise a significant majority in the South, than the higher rates of intervention there are related to population differences, not to differences in
management among that population - so understanding these differences as a percentage of population differences within the region would be helpful
3) There have been changes in practice and outcomes between 2006-2017 - I would have been interested to know if there were differences in statistics/ outcomes between the first half of this time range and the 2nd half
4) is there any data on varying religious affiliations? If people are more religious in the South, this could account for differences in counseling and desire for intervention that has nothing to do with race
5) what about the outcome statistics - if these were collected out to 1 year, why is outcome on day 1 all that is presented?
I would be curious to see if the more aggressive interventions resulted in differences in survival/outcome beyond just 1 day
Reviewer #3:
Overview: the rationale for this study stems from previous studies suggesting that minority peri viable infants receive more intensive treatment than weight infants. However these prior studies evaluated the entire United States not
subsided by regions. This is a 11 year cohrt study to evaluate periviable (22-25 weeks) outcomes. The primary outcome evaluated were antenatal corticosteroid use, cesarean delivery, and postnatal support. This is a national study based upon the Vermont Oxford Network data. The United States was divided into 4 regions ( northeast, Midwest, West, South ) and 4 race/ethnicity (White, Black, Hispanic, Asian). The authors concluded that there is both regional and racial differences in terms of outcomes and potential management of periviable fetus.
Summary of Comments: Overall this is an important topic that the authors had a clear plan with collecting and analyzing the data. The literature can benefit from the results of such a large dataset. The specific aims of the manuscript were achieved. However, the manuscript in its current form is not easy to read because of structure and graphics. Also, there needs to be more methodology details and consolidation of graphs with the addition of numerators/denominators so that readers can follow along easily.
Comments:
Give example of the hospitals that fall under the "other" category of hospital ownership type. Also since it is such a small percentage the "other" and "investor-owned "can be combined into one category.
Explain obstetrical level classification and consider combining level 1 and level 2
Caesarean delivery should be categorized as primary or repeat. If unable to do this then it should be mentioned as a limitation in the discussion since higher amount of C-sections are regional and some women may have been undergoing a repeat C-section.
Since there is such a robust data set, the data would have been much cleaner if containing only be years in which she had complete data ( i.e. the maternal hypertension, chorio).
Chorio was never mentioned in the methods. With this included in the modeling? If not it should be because it plays a critical role in decision making for management of periviable infants.
eTable 1 is vitals for reader to know your population/demographics. Therefore it should be the 1st table, not an electronic table. There should also be a final column that tallies the rows.
Starting on line 176 - instead of listing the percentages already in table 1 a more narrative description of the study sample would be informative (ie there was a 47.8% increase in antenatal steroid from 22-23 week )
Figures 1-3 - I appreciate that the figure shows the great variation between hospitals in each region. However the statement that all 3 variables are increased in the Midwest and South for 22-23 weeks is not supported by the data presented. In figure 1 the confidence interval at 23 weeks for the North East in the South overlap. Furthermore it would still be useful to have a table or next to the percentages the raw numbers for each region versus just having the
percentages listed.
Throughout the manuscript the numerator and denomentor for the percentages need to be listed
eTable2-5: the number (N) for the outcomes need to be listed, not just number of observations in the model. Also, rationale/procedure used to choose reference variables needs to be explained in methods and in legend.
eTable 3 lists maternal HTN as the REF but then again with an RR. Was the REF used in the model "no HTN" and those with HTN were at and increases risk of CD in all regions? The HTN results mentioned in the "discussion" but the significant result should be stated in the results as well.
For the primary outcome depicted in eFig 4-6, should be displayed graphically in a more condensed manner
Figures 1-3 could benefit from color coding by hospital type and/or NICU for to purpose of expounding on prior studies that are mentioned in the discussion.
For limitations, not listing if CD was repeat or primary as this can affect the conclusions
STATISTICAL EDITOR COMMENTS:
The Statistical Editor makes the following points that need to be addressed:
lines 161-165: The aRR rates should be also be adjusted for cluster analysis based on the hospital unit. It is not clear from the language used to describe the methodology whether that was specifically addressed. There may have been strong clustering of protocols or guidelines within an individual hospital that then led to implementation of ANS, CD etc. That is, the Hospital characteristics cited in Table 1 ( NICU type, OB level, hospital ownership) are group characteristics, not at the level of the individual 776 NICUs in the VON).
Another major issue is the number of comparisons by GA(4), region(4), race/ethnicity(3 vs referent) for 3 separate outcomes of interest. That is, a total of 144 combinations. Using the standard P < .05 for inference testing will lead to many spurious associations. The large sample sizes (without adjustment for clustering) raises the potential for statistically significant, but clinically minor, associations. Need to use a stricter inference threshold, in accordance with multiple
hypothesis testing.
General: Need to enumerate all missing data, particularly for proportion missing for race/ethnicity and if any missing data were selectively allocated by region or other stratifier. This will help the reader to understand how generalizable the data may be.
lines 331-337, 342-344: True, but there is data from individual NICUs, so that any difference by cluster could be evaluated.
EDITOR COMMENTS:
1. Specific manuscript comments:
We no longer require that authors adhere to the Green Journal format with the first submission of their papers. However, any revisions must do so. I strongly encourage you to read the instructions for authors (the general bits as well as those specific to the feature-type you are submitting). The instructions provide guidance regarding formatting, word and reference limits, authorship issues, and other things. Adherence to these requirements with your revision will avoid delays during the revision process, as well as avoid re-revisions on your part in order to comply with the formatting. Pay
particular attention to information about abbreviations, for example.
Please clearly state a hypothesis, primary and secondary outcomes. Then report your results and organize your discussion addressing your primary and then your secondary outcomes.
PRESENTATION OF STATS INFORMATION (P Values vs Effect Size and Confidence Intervals)
While P values are a central part of inference testing in statistics, when cited alone, often the strength of the conclusion can be misunderstood. Whenever possible, the preferred citation should be in terms of an effect size, such as odds ratio or relative risk or the mean difference of a variable between two groups, expressed with appropriate confidence intervals.
When such syntax is used, the P value has only secondary importance and often can be omitted or noted as footnotes in a Table format. Putting the results in the form of an effect size makes the result of the statistical test more clinically relevant and gives better context than citing P values alone.
This is true for the abstract as well as the manuscript.
Please provide absolute values for variables, in addition to assessment of statistical significance. For instance, in the results section of the abstract, there should be data provided (as described above) to support the findings, rather than simply stating the findings.
Please note that effect sizes (RR, OR) within the zone of potential bias should be noted as weak. Those effect sizes in the zone of potential interest should be emphasized. (Ref: False alarms and pseudo-epidemics. The limitations of observational epidemiology. Grimes DA, Schulz KF. Ob Gyn 2012;120:920-7) The very small point estimates for some of your data needs to be addressed in your discussion section.
We ask that you provide crude OR’s followed by adjusted OR’s for all variables.
Line 62: Your precis indicates that you are studying 22-23 week EGA pregnancies so this reference to 22-25 weeks seems odd. Your title includes “periviable” and 25 weeks doesn’t really seem to fall into that range any more. Please comment.
Line 63: perhaps “maternal administration of corticosteroids” instead of “provision of antenatal corticosteroids”
Line 68: rather than say “We report” please just report the findings.
Line 112: are you suggesting the previously identified differences in rates of active intervention may due to provider differences by region or that the differences identified before may vary be region due to population density of minority families in different regions? Seems that either could be true (or some other explanations). Could you edit for clarify?
116: Its’ not racial/ethnic differences in the provision of these services, its in the receipt of these services. The clinicians provide the services, the patients receipt them
138 Has VON not adopted the Levels of Maternal Care from several years ago which are now in 4 levels? As the obstetric levels you describe are not currently used, it would be best if you define them.
Line 141: Current terminology is hypertensive disorders of pregnancy. Please edit. Your descriptions of BP don’t really coordinate with Obstetrical descriptions. Edema is no longer considered part of the diagnostic criteria for preeclampsia; I am unclear what you mean by BP>140 systolic or diastolic 90 before or during the present rpregnancy. It would seem these would fall within “chronic” or “hypertensive disorders of pregnancy”. Since this is an OB journal, these definitions require some clarity as they don’t align with ob practice.
Line 151: Note statistical editors requested changes in your analysis.
172: once you define your primary and secondary outcomes, please make sure that you organize your results and discussion w/ primary followed by secondary outcomes.
Lines 176-178: are there statistically significant differences between any of these results?
Line 189: How is it known that it is is the increased rate of pediatric interventions, and not the OB interventions, that “are reflected in survival rates to day 1”? Perhaps it would be more accurate to say that increased rates of active intervention (Steroids, CS, and PLS or some combination thereof) that is reflected in the survival rates?
Line 188: Not sure this note belongs on this specific line but it needs to be addressed somewhere. Without knowing how many fetuses were alive prior to labor or prior to delivery (ie, how many MAY potentially have benefitted from cesarean to prevent a stillbirth, its impossible to really know what intervention contributes to survival. Those infants that were stillborn and delivered by CS would not show up in your “neonatal survival data” since they were not alive at birth.
Line 272: could you suggest some other possible explanations besides lack of guidelines? Population differences perhaps?
283; not sure you know these areas are more “restrictive” which sounds like something is imposed. Perhaps, more conservative in their use.
This gets a bit to what may be a different perspective between pediatricians and obstetricians, which plays out at this periviable edge. With the poor long term outcomes for the periviable fetus that most obstetricians assume, they may be less willing to use cesarean, for instance, which would almost certainly result in a classical CS with significant long term
(and short term) implications, including all future deliveries by CS and in the later preterm period to avoid uterine rupture.
Line 278: Please use reVITALize terminology—now called preterm prelabor rupture of the membranes.
Line 311: why just call out PLS? May be true for steroids, CS as well.
Line 331: This makes sense—hypertension is more often at this gestational age an INDICATION for delivery and Cesarean rather than induction of labor much more likely in the periviable pregnancy with such severe hypertensive disease. Most of the hypertensive mothers of periviable neonates will not be in labor so their births will be iatrogenic—and CS much more likely.
For all tables: please make sure you provide the absolute numbers and not just the point estimats and CI's in all of your tables.
We don’t use a designation of “eTable” by which I assume you mean available on line only. Please move all of this data to be within the manuscript, not just on line.
2. The Editors of Obstetrics & Gynecology are seeking to increase transparency around its peer-review process, in line with efforts to do so in international biomedical peer review publishing. If your article is accepted, we will be posting this revision letter as supplemental digital content to the published article online. Additionally, unless you choose to opt out, we will also be including your point-by-point response to the revision letter. If you opt out of including your response, only the revision letter will be posted. Please reply to this letter with one of two responses:
A. OPT-IN: Yes, please publish my point-by-point response letter.
B. OPT-OUT: No, please do not publish my point-by-point response letter.
3. As of December 17, 2018, Obstetrics & Gynecology has implemented an "electronic Copyright Transfer Agreement"
(eCTA) and will no longer be collecting author agreement forms. When you are ready to revise your manuscript, you will be prompted in Editorial Manager (EM) to click on "Revise Submission." Doing so will launch the resubmission process, and you will be walked through the various questions that comprise the eCTA. Each of your coauthors will receive an email from the system requesting that they review and electronically sign the eCTA.
Please check with your coauthors to confirm that the disclosures listed in their eCTA forms are correctly disclosed on the manuscript's title page.
4. In order for an administrative database study to be considered for publication in Obstetrics & Gynecology, the database used must be shown to be reliable and validated. In your response, please tell us who entered the data and how the accuracy of the database was validated. This same information should be included in the Materials and Methods section of the manuscript.
5. All studies should follow the principles set forth in the Helsinki Declaration of 1975, as revised in 2013, and manuscripts should be approved by the necessary authority before submission. Applicable original research studies should be reviewed by an institutional review board (IRB) or ethics committee. This review should be documented in your cover letter as well in the Materials and Methods section, with an explanation if the study was considered exempt. If your research is based on a publicly available data set approved by your IRB for exemption, please provide documentation of this in your cover letter by submitting the URL of the IRB website outlining the exempt data sets or a letter from a representative of the IRB. In addition, insert a sentence in the Materials and Methods section stating that the study was approved or exempt from approval. In all cases, the complete name of the IRB should be provided in the manuscript.
6. Standard obstetric and gynecology data definitions have been developed through the reVITALize initiative, which was convened by the American College of Obstetricians and Gynecologists and the members of the Women's Health Registry Alliance. Obstetrics & Gynecology has adopted the use of the reVITALize definitions. Please access the obstetric and gynecology data definitions at https://www.acog.org/About-ACOG/ACOG-Departments/Patient-Safety-and-Quality- Improvement/reVITALize. If use of the reVITALize definitions is problematic, please discuss this in your point-by-point response to this letter.
7. Because of space limitations, it is important that your revised manuscript adhere to the following length restrictions by manuscript type: Original Research reports should not exceed 22 typed, double-spaced pages (5,500 words). Stated page limits include all numbered pages in a manuscript (i.e., title page, précis, abstract, text, references, tables, boxes, figure legends, and print appendixes) but exclude references.
8. Specific rules govern the use of acknowledgments in the journal. Please note the following guidelines:
* All financial support of the study must be acknowledged.
* Any and all manuscript preparation assistance, including but not limited to topic development, data collection, analysis, writing, or editorial assistance, must be disclosed in the acknowledgments. Such acknowledgments must identify the entities that provided and paid for this assistance, whether directly or indirectly.
* All persons who contributed to the work reported in the manuscript, but not sufficiently to be authors, must be
acknowledged. Written permission must be obtained from all individuals named in the acknowledgments, as readers may
infer their endorsement of the data and conclusions. Please note that your response in the journal's electronic author form verifies that permission has been obtained from all named persons.
* If all or part of the paper was presented at the Annual Clinical and Scientific Meeting of the American College of Obstetricians and Gynecologists or at any other organizational meeting, that presentation should be noted (include the exact dates and location of the meeting).
9. The most common deficiency in revised manuscripts involves the abstract. Be sure there are no inconsistencies between the Abstract and the manuscript, and that the Abstract has a clear conclusion statement based on the results found in the paper. Make sure that the abstract does not contain information that does not appear in the body text. If you submit a revision, please check the abstract carefully.
In addition, the abstract length should follow journal guidelines. The word limits for different article types are as follows:
Original Research articles, 300 words. Please provide a word count.
10. Only standard abbreviations and acronyms are allowed. A selected list is available online at http://edmgr.ovid.com /ong/accounts/abbreviations.pdf. Abbreviations and acronyms cannot be used in the title or précis. Abbreviations and acronyms must be spelled out the first time they are used in the abstract and again in the body of the manuscript.
11. The journal does not use the virgule symbol (/) in sentences with words. Please rephrase your text to avoid using
"and/or," or similar constructions throughout the text. You may retain this symbol if you are using it to express data or a measurement.
12. In your Abstract, manuscript Results sections, and tables, the preferred citation should be in terms of an effect size, such as odds ratio or relative risk or the mean difference of a variable between two groups, expressed with appropriate confidence intervals. When such syntax is used, the P value has only secondary importance and often can be omitted or noted as footnotes in a Table format. Putting the results in the form of an effect size makes the result of the statistical test more clinically relevant and gives better context than citing P values alone.
If appropriate, please include number needed to treat for benefits (NNTb) or harm (NNTh). When comparing two procedures, please express the outcome of the comparison in U.S. dollar amounts.
Please standardize the presentation of your data throughout the manuscript submission. For P values, do not exceed three decimal places (for example, "P = .001"). For percentages, do not exceed one decimal place (for example, 11.1%").
13. Please review the journal's Table Checklist to make sure that your tables conform to journal style. The Table Checklist is available online here: http://edmgr.ovid.com/ong/accounts/table_checklist.pdf.
14. The American College of Obstetricians and Gynecologists' (ACOG) documents are frequently updated. These
documents may be withdrawn and replaced with newer, revised versions. If you cite ACOG documents in your manuscript, be sure the reference you are citing is still current and available. If the reference you are citing has been updated (ie, replaced by a newer version), please ensure that the new version supports whatever statement you are making in your manuscript and then update your reference list accordingly (exceptions could include manuscripts that address items of historical interest). If the reference you are citing has been withdrawn with no clear replacement, please contact the editorial office for assistance ([email protected]). In most cases, if an ACOG document has been withdrawn, it should not be referenced in your manuscript (exceptions could include manuscripts that address items of historical interest). All ACOG documents (eg, Committee Opinions and Practice Bulletins) may be found via the Clinical Guidance &
Publications page at https://www.acog.org/Clinical-Guidance-and-Publications/Search-Clinical-Guidance.
15. Figures 1–4: Please upload as high-res figure files (eps, tiff, jpeg, etc.) on Editorial Manager.
When you submit your revision, art saved in a digital format should accompany it. If your figure was created in Microsoft Word, Microsoft Excel, or Microsoft PowerPoint formats, please submit your original source file. Image files should not be copied and pasted into Microsoft Word or Microsoft PowerPoint.
Please upload each figure as a separate file to Editorial Manager (do not embed the figure in your manuscript file).
16. Authors whose manuscripts have been accepted for publication have the option to pay an article processing charge and publish open access. With this choice, articles are made freely available online immediately upon publication. An
information sheet is available at http://links.lww.com/LWW-ES/A48. The cost for publishing an article as open access can be found at http://edmgr.ovid.com/acd/accounts/ifauth.htm.
Please note that if your article is accepted, you will receive an email from the editorial office asking you to choose a publication route (traditional or open access). Please keep an eye out for that future email and be sure to respond to it promptly.
17. If you choose to revise your manuscript, please submit your revision through Editorial Manager at
http://ong.editorialmanager.com. Your manuscript should be uploaded in a word processing format such as Microsoft Word.
Your revision's cover letter should include the following:
* A confirmation that you have read the Instructions for Authors (http://edmgr.ovid.com/ong/accounts/authors.pdf), and
* A point-by-point response to each of the received comments in this letter.
If you submit a revision, we will assume that it has been developed in consultation with your co-authors and that each author has given approval to the final form of the revision.
Again, your paper will be maintained in active status for 14 days from the date of this letter. If we have not heard from you by Nov 15, 2019, we will assume you wish to withdraw the manuscript from further consideration.
Sincerely,
Nancy C. Chescheir, MD Editor-in-Chief
2018 IMPACT FACTOR: 4.965
2018 IMPACT FACTOR RANKING: 7th out of 83 ob/gyn journals __________________________________________________
In compliance with data protection regulations, you may request that we remove your personal registration details at any time. (Use the following URL: https://www.editorialmanager.com/ong/login.asp?a=r). Please contact the publication office if you have any questions.
Date: Dec 10, 2019 To: "Nansi Boghossian"
From: "The Green Journal" [email protected] Subject: Your Submission ONG-19-1890R1
RE: Manuscript Number ONG-19-1890R1
Regional and Racial/Ethnic Differences in Perinatal Interventions among Periviable Births Dear Dr. Boghossian:
Your revised manuscript has been reviewed by the Statistical Editor. We are requesting that you make additional edits to your submission prior to the Editors making a final decision.
Each point raised requires a response, by either revising your manuscript or making a clear and convincing argument as to why no revision is needed. To facilitate our review, we prefer that the cover letter include the comments made by the reviewers and the editor followed by your response. The revised manuscript should indicate the position of all changes made. We suggest that you use the "track changes" feature in your word processing software to do so (rather than strikethrough or underline formatting).
Your paper will be maintained in active status for 14 days from the date of this letter. If we have not heard from you by Dec 24, 2019, we will assume you wish to withdraw the manuscript from further consideration.
STATISTICAL EDITOR:
I accept the Authors' responses to most of my questions and thank them for the additional efforts. However, their response regarding multiple hypothesis testing (specifically in the R1 version Table 2 of adjusted and unadjusted RRs) is unacceptable.
The Authors claim in Precis, Abstract and in main text that their study shows that "major regional and racial/ethnic differences exist in receiving antenatal steroids, CD and post-natal life support". These statements imply that this is not meant to be an epidemiologic survey, looking for possible hypotheses, but rather that conclusions are being made regarding the relationship among race/ethnicity, geographic region and the previously mentioned medical care.
The Authors need to balance the need for identifying non-spurious, generalizable associations with the loss of statistical power consequent to stricter inference thresholds. Most of the associations shown in Table 2 have only modest strength (RRs and aRRs in the 0.80-1.20 range) and many of the CIs boundaries are very close to 1.00. Using even a small incremental change to 0.01 threshold or CIs of 99th %-tile rather than 95% %-tile would render many of the associations no longer significant.
The Authors need to both change the inference threshold to at least 0.01, rather than 0.05 and to include among limitations in DIscussion that testing multiple hypotheses with application of the standard .05 threshold runs the risk of accepting spurious associations.
EDITOR'S COMMENTS:
Thank you for your revised submission to Obstetrics & Gynecology. In addition to the comments from the Statistical Editor above, you are being sent a notated PDF that contains the Editor’s specific comments. Please review and consider the comments in this file prior to submitting your revised manuscript. These comments should be included in your point-by- point response cover letter.
***The notated PDF is uploaded to this submission's record in Editorial Manager. If you cannot locate the file, contact Randi Zung and she will send it by email - [email protected].***
1. I respect that you prefer to avoid multiple comparison testing: there is a wide divergence among biostatisticians on this point. Having said that, the Journal's standard is to require more conservative thresholds for significance in research with large numbers of comparisons, such as yours.
2. Please include the results of the analysis you provided in response to Reviewer 2 which looks at Race/Ethnicity across all 4 regions. This reviewer asked a significant question and while you can make the disclaimer that you make about effect modifier due to region in the discussion, the data is relevant to understanding this very complex problem.
3. In the section "Regional Variation in Survival," please add the data about survival to 49 days which you provide above.
4. Through your submission, please use "associations," not "predictors."
5. In your Discussion where you start with, "Our study expands on these findings by describing...": I want to push you a bit on this statement that you are describing only physician practice. While literally true, this practice is not performed in a vacuum and particularly at 22-23 weeks, when the proactive management decisions are based on maternal decision making to a large degree, what you are really describing is the end result of physician and patient decision making. This is supported by the loss of much difference at 24-25 weeks, at a gestational age at which there may be fewer ways that the mother can intervene re: post natal resuscitation. You don't include this incredibly important influence in you sentence starting with "at this threshold of viability"....
Sincerely,
Nancy C. Chescheir, MD Editor-in-Chief
__________________________________________________
In compliance with data protection regulations, you may request that we remove your personal registration details at any time. (Use the following URL: https://www.editorialmanager.com/ong/login.asp?a=r). Please contact the publication office if you have any questions.
