The district is a rural area with a population of 4.1 million, characterized by homogeneity in terms of ethnicity, culture,

and language (UNICEF, 2008). Ethical approval for the study was received from The Bangladesh Medical Research Council (Ref. no. BMRC/Eth.C/2008/402) in Bangladesh and the Regional Ethical Review Board at Karolinska Institutet, Stockholm, Sweden (Ref. no. 2008/919-31). Participants and procedures Participants for the study were selected from a sample of mothers identified as having depressive symptoms 2–3 months postpartum in the longitudinal study, that is, having a score of 10 or more on the Edinburgh Postnatal Depressive Scale (EPDS; Cox, Rucaparib Holden, & Sagovsky, 1987). The mothers were approached consecutively by a trained female Bangladeshi anthropologist, informed about the study, and asked if they wanted to participate in an open narrative interview. Twenty-one mothers of infants (both primi- and multiparas)— whose infants were between 3 and 9 months old (M=6.2 months) were approached, informed, and agreed to participate. The open narrative

interviews of approximately 2 h were conducted between January and July 2009 in Bangla by the trained Bangladeshi anthropologist at the women’s homes. Except for some background data, the interviews were loosely structured and the interviewer asked the woman to narrate her experiences and concerns after childbirth. If the mother had difficulties narrating freely, then she was asked specifically about the delivery; the postpartum period; social support; PI3K inhibitor breastfeeding; relationship to the infant, partner, and family; and health in the family. All interviews were found tape-recorded and transcribed verbatim in Bangla and translated into English prior to coding. One of the co-authors (ZNK) who is bilingual read all the interviews in Bangla and controlled the English translation against the Bangla version. In case of a discrepancy in

the translation between Bangla and English, it was discussed between the two bilingual authors (ZKN and HEN). Data analysis The data were analysed using inductive content analysis (Granehielm & Lundman, 2004). Content analysis is a research technique for making valid inferences from texts to the contexts of their use (Krippendorff, 2013). The first author (ME) read through the English translations of the interviews several times to obtain a sense of the whole. One bilingual researcher (ZKN) read all the interviews in Bangla and another (HEN) read half of the interviews. For each transcript, the first author (ME) identified the meaning units related to the aim, marked them, and condensed the meaning units. The condensed meaning units were then abstracted and labelled with a code according to Granehielm and Lundman (2004). To increase the credibility of the analysis, a bilingual anthropologist and the first author (ME) independently coded four interviews in English.