Title: Socioeconomic inequalities in stillbirth rates in Europe: measuring the gap using routine data from the Euro-Peristat Project

Transcription

1 Author s response to reviews Title: Socioeconomic inequalities in stillbirth rates in Europe: measuring the gap using routine data from the Euro-Peristat Project Authors: Jennifer Zeitlin (jennifer.zeitlin@inserm.fr) Laust Mortensen (lamo@sund.ku.dk) Caroline Prunet (caroline.prunet@inserm.fr) Alison Macfarlane (A.J.Macfarlane@city.ac.uk) Ashna Hindori-Mohangoo (ashna.mohangoo@tno.nl) Mika Gissler (mika.gissler@thl.fi) Katarzyna Szamotulska (szamotul@imid.med.pl) Karin van der Pal (karin.vanderpal@tno.nl) Francisco Bolumar (francisco.bolumar@uah.es) Anne-Marie Andersen (amny@sund.ku.dk) Helga Sol Olafsdottir (helgasol@landspitali.is) Wei-Hong Zhang (wzhang@ulb.ac.be) Béatrice Blondel (beatrice.blondel@inserm.fr) Sophie Alexander (salexand@ulb.ac.be) Version: 1 Date: 23 Nov 2015 Author s response to reviews: Dear Editor,

2 Thank you for giving us the opportunity to revise our manuscript; we are also grateful to the reviewers for their helpful comments and suggestions for improvement. We explain below how we have integrated these into the revised version (page numbers refer to marked version). Reviewer #1 This study is able to derive important messages despite the obvious challenges of available data, due to incomplete recording of social and educational levels, and differing definitions of stillbirth. The epidemiological or public health related observations could be made more relevant for clinical readers. The 25% potential savings in stillbirths if all mothers had the same high educational or occupational level will not happen in the immediate future. Evidence is alluded to that social and educational factors represent a number of more direct risk factors such as obesity and smoking. But what proportion of the excess risk do they account for - and after they are considered, does social gradient remain a risk? Are we missing important socio-economic related variables? A study which attempted to deal with this [1], but is not mentioned, found increased stillbirth risk associated with area based social deprivation (lowest IMD quintile: RR 1.6) and maternal (RR 1.2) and paternal (RR 1.3) unemployment - but in the univariate analysis only. The significance disappeared in the multivariable analysis, which identified hitherto unknown or unmeasured 'attributable risks' associated with several potentially modifiable risk factors, especially unrecognised fetal growth restriction. Discussion of such observations would firstly help to highlight the need to see social factors as potential warning signs in clinical care, and secondly, help strengthen the call for better routine data collection to improve our understanding of social gradient related variation in stillbirth risk. Response: We are in full agreement with the reviewer about the multiple pathways that link educational achievement to poor outcomes, including lower socioeconomic status, (un)healthy behaviors, diet, smoking, poor access to care The question of how much these pathways contribute to explaining the impact of education on perinatal outcomes is complex and the current literature is not concordant (some of the studies find that the social gradient disappears, while others do not). We believe this is an important area for further research, especially as regards those factors that are potentially modifiable, as the reviewer points out. We also acknowledge that there

3 may be a pathway through access to care and screening for FGR and have added the following text. Antenatal detection of growth restriction may play a role in preventing stillbirth and the extent to which social position affects access to screening should also be considered [37]. The question of how much of the social gradient is explained by these behavioral and healthcare factors, especially those that are potentially modifiable, is an important area for further research.. Minor points: 1. Page 11, line 21: The reference category is presumably again mothers with post-secondary education here, and should be repeated for clarity. Response: we have rephrased this to improve clarity. 2. Abstract - Results: the RR values are quoted wrongly, i.e. in reverse: to give RRs >1, it should read 'women with primary and lower secondary education.compared to post-secondary education' etc. Response: thank you for noticing this error which has been corrected. Reference 1. Gardosi J, Madurasinghe V, Williams M, Malik A, Francis F. Maternal and fetal risk factors for stillbirth: population based study. BMJ 2013;346:f Reviewer #2 This interesting paper presents a detailed analysis of stillbirth rates according to different socioeconomic groups amongst countries in Europe. The paper is clear and well written and I have only a few points of correction/clarification.

4 Major revisions 1. My main concern about this paper is the missing data on educational and socioeconomic groups. The authors make little comment on this other than to list the absolute percentages of missing data. I wondered whether these data might be differentially missing i.e. that the educational and/or socioeconomic group of parents of stillborn babies are fully classified whereas substantial numbers of live births have missing data. Many previous studies have suggested that such data are not missing at random and more likely to be missing from lower socio-economic groups. This would effectively reduce the size of the denominator group for those of lower socioeconomic status and could thus be partially responsible for the effects seen. Could the authors comment and/or clarify whether data were missing equally amongst live and stillbirths? Response: We agree with the reviewer that the differences in missing data raise questions for this analysis. Missing data were more frequent for the stillbirths than live births. We have added more analysis and information to the manuscript about the missing data (in the methods, results and the discussion sections). We also correlated the proportion of missing data in each country with our outcome measures to test whether countries with more missing data have attenuated or more pronounced social gradients. These correlations are not significant (null for education, slightly negative for occupation). We also added this point to the study limitations and cite one study which examined the characteristics of women with missing birth certificate data in Belgium. Please see changes on pages: 8, 9, 10, 11 and The distributions across groups between countries were extremely variable and to me this raises major concerns about the comparability of the classifications. Can the authors comment on whether they made any efforts to validate their groupings and/or checked through other sources that these did indeed represent the patterns of educational and socio-economic distributions within those countries? Response: We have added more information about this and cited two recent studies on educational inequalities in preterm birth that used birth cohort data and also found large differences in the distribution of maternal education, as follows: "There are also real differences in educational attainment across Europe. Similarly broad variations in maternal education have been documented in comparative studies based on birth

5 cohorts across Europe [27, 28]. However, despite the differences in the distribution of maternal educational attainment, a negative social gradient was of stillbirth rates observed in most countries when this indicator was used." Minor revisions 1. The abstract to me appears to be written the wrong way round i.e. it appears to read that women from a higher educational status have a higher relative risk of stillbirth and I think the authors should reverse the phrasing, for example: 'The median RR of stillbirth for women with primary and lower secondary education compared to women with post-secondary education was 1.9 ' Response: thank you for noticing this error which has been corrected. 2. Can the authors clarify for readers from outside the UK why there are apparently no unemployed or students in England and Wales (because unemployed fathers are classified according to their most recent occupation)? Response: this classification does not include a category for unemployed (as the classification should be according to the most recent occupation, as noted). Students were included in the unclassified category (we would have had to request a separate analysis to distinguish these cases) and therefore were considered to be missing. Editorial Requests If your study involves humans, human data or animals, then your article should contain an ethics statement which includes the name of the committee that approved your study. If ethics was not required for your study, then this should be clearly stated and a rationale provided. Response: Euro-Peristat uses aggregate routinely collected data transferred following national legislation and therefore ethics approval is not required. We have added a sentence to the methods about the need for ethical approval: This study is based on aggregated routinely collected data, so ethics approval was not required.

6 Consent: If your article is a prospective study involving human participants then your article should include a statement detailing consent for participation. If individual clinical data is presented in your article, then you must clarify whether consent for publication of these data was obtained. Response: Not applicable Availability of supporting data: BioMed Central strongly encourages all data sets on which the conclusions of the paper rely be either deposited in publicly available repositories (where available and appropriate) or presented in the main papers or additional supporting files, in machine-readable format whenever possible. Authors must include an Availability of Data and Materials section in their article detailing where the data supporting their findings can be found. The Accession Numbers of any nucleic acid sequences, protein sequences or atomic coordinates cited in the manuscript must be provided and include the corresponding database name. Response: We have added an Availability of Data and Materials section: data used in this analysis can be requested from the Euro-Peristat coordination team through the project s website. Authors Contributions: Your 'Authors Contributions' section must detail the individual contribution for each individual author listed on your manuscript. Response: We have added the initials of the members of the group author list.