Responses
Other responses
Jump to comment:
- Published on: 17 May 2017
- Published on: 24 March 2017
- Published on: 17 May 2017RE: Author's response
We thank Cooney and Partridge for their interest in our article.1 We believe, however, that they have misunderstood much of what we wrote. We accessed websites with publicly available national and subnational data on population-level rates of religious affiliation and breastfeeding initiation. As we explained, two other selection criteria reduced the number of countries we examined: (1) sufficient within-population variability in both religious affiliation and breastfeeding initiation and (2) data on both variables, so that they could be associated at the same ecological (geographical) level. As we acknowledged, we did not find Lander-level data for Germany, the country that best met our selection criteria. Our inclusion of data from French-speaking regions is clearly a strength of our study--not a limitation. We used Protestant affiliation as defined in the databases we found, and we focused on the major Protestant groups in the general population. Since the data we used are publicly available, we invite Cooney and Partridge to analyse the associations between minor Protestant groups and breastfeeding rates in Ireland. Regarding the definition of Protestants, any degree of misclassification should have reduced, rather than increased, the associations we observed. Our scatterplots show the crude (unadjusted) data together with the crude correlations. For France, Ireland, the UK and Canada, crude and adjusted results were all in the same direction: a negative correlation be...
Show MoreConflict of Interest:
None declared. - Published on: 24 March 2017RE: Concerns about a recent study on the influence of religion on breastfeeding ratesWe are concerned that the recent paper by Bernard et al published in BMJ Global Health 2016 fails to provide a scientific basis to support the stated conclusion that within Western countries the proportion of Catholics in the population influences the breastfeeding initiation rates.The paper does not include a literature review on the role of religion in relation to infant feeding and there is no convincing scientific rationale presented to support the hypothesis being tested. There then follows a description of the study’s methods and results which reveals a series of methodological and reporting problems which, in our view, renders the work wide open to biases, confounders and incorrect deductions.There is evidence of selection bias in the five countries selected for within-country analyses, all limited to specific Western countries, with no explanation as why these particular countries were selected for inclusion. It is noteworthy that of these selected five, the dominant languages for each is either English or French which suggests language bias may also be a factor.A further problem with the methods is the unexplained differences in the selection of the Protestant populations for each of the five countries. For the US the researchers included two categories of Protestants in their analysis (Evangelical Protestants and Mainline Protestants), for Canada and France the category of Protestants was used but then for the U...Show MoreConflict of Interest:
None declared.