**2. Experimental Section**

#### *2.1. Recruitment of Subjects*

A prospective cohort study of infant feeding practices among women in Kuwait was conducted between October 2007 and October 2008. The study methods have been described previously [30], but briefly mothers were recruited from three major public hospitals and one private hospital located in Kuwait City. Within 72 h of delivery, eligible mothers were visited and invited to participate in the study by the researcher (MD) who provided a written and verbal description of the study. Women were considered to be eligible for the study if they were able to read or understand Arabic or English and had delivered a live, healthy, singleton of 36 weeks or more gestational age. Mothers whose infants were admitted to the Special Care Nursery (SCN) for minor illnesses or observation were eligible for recruitment.

The study was approved by the Medical Faculty Ethics Committee of the University of Glasgow, (Application No. FM03906: Approved May 29, 2007) and by the Ministry of Health in Kuwait. Participants provided signed informed consent and were advised that they could refuse to participate or withdraw from the study at any time, without prejudicing their post-natal care or the care of their baby.

#### *2.2. Data Collection*

Mothers who agreed to join the study were interviewed face-to-face to complete a baseline questionnaire prior to discharge from hospital. Women who declined to participate were asked to provide some basic socio-demographic data to determine if the sample were representative of the population of women giving birth at the participating hospitals. All participants were followed up by telephone interview at 6, 12, 18 and 26 weeks postpartum. Data were collected using questionnaires previously used in similarly designed studies of Australian women [31] and modified slightly to meet the needs of this study population. Information on socio-demographic characteristics, maternal lifestyle factors, infant characteristics, biomedical factors, hospital practices, psychosocial factors and feeding practices were collected at baseline. Information on current feeding practices, changes to feeding practices, breastfeeding experiences and introduction of a pacifier were collected during follow-up interviews.

#### Infant Feeding Assessments

Breastfeeding terms used in this study were those defined by the World Health Organization [3]. An infant was considered to be *exclusively breastfed* when he or she had received only breast milk with no other liquids (except for oral rehydration solutions (ORS), drops or syrups) or complementary (solid) foods, and to be *predominantly breastfed* when he or she received breast milk as the main source of nourishment, with certain liquids (water, water-based fluids, fruit juices, ritual fluids, ORS, drops or syrups) but received no other liquids (including formula milk and non-human milks) or complementary foods. *Full breastfeeding* was defined as either exclusive or predominant breastfeeding [32]. *Any breastfeeding* was defined as an infant who was receiving breast milk, with or without formula, other milks, fluids or complementary foods. Duration of exclusive, predominant and any breastfeeding was determined by using information about the age at which other types of milks, liquids (e.g., water, fruit juice) and/or complementary foods were introduced in the first six months of life. In this study, *prelacteal feeding* was defined as the act of giving any liquid or food item (except breast milk) to a newborn within the first three days after birth [33].

#### *2.3. Statistical Analysis*

We explored the associations of breastfeeding duration and a variety of characteristics and practices reported in the literature to be associated with the duration of breastfeeding using Cox's proportional hazards model. This model allows joint estimation of the effects of independent variables on the risk of cessation of breastfeeding and can be used to analyze data that contain censored observations [34]. We tested the role of: (1) socio-demographic factors (maternal age, education, country of birth, employment plans for six month postpartum); (2) maternal lifestyle factors (pre-pregnancy BMI calculated from self-reported weight and height, smoking during pregnancy); (3) infant factors (gender, having spent time in the SCN); (4) biomedical factors (parity, delivery method, breastfeeding problems at baseline, breastfeeding problems at six weeks postpartum, age at which pacifier was introduced) (5) hospital practices (time to first breastfeed, composition of infant's first feed, use of prelacteal feeds, infant roomed-in, infant demand fed in hospital); and (6) psychosocial factors (when decided on feeding method, whether pregnancy was planned, father's feeding preference, maternal grandmother's feeding preference, paternal grandmother's feeding preference). The association of individual variables and the duration of any and full breastfeeding was first evaluated in a univariate model. Any variable with a *P*-value of <0.100 was then included in a multivariate model which was reduced using the backward stepwise procedure. The fitness of each model was assessed at every step to avoid dropping non-significant variables that affected the model fitness. All variables in the final model were variables for which, when excluded, the change in deviance compared with the corresponding statistics on the relevant degrees of freedom was significant. Statistical analyses were performed using the Statistical Package of Social Sciences version 19 (SPSS Inc., Chicago, IL, USA).

#### **3. Results**

A total of 439 women were invited to participate in the study and 373 mothers completed the baseline questionnaire while in hospital, giving a response rate of 85%. There were no significant differences between participants and those declining to participate (*n* = 66) with respect to age (Ȥ 2 4.413, *P* = 0.110), level of education (Ȥ 2 2.455, *P* = 0.117) and chosen method of feeding at discharge (Ȥ 2 447, *P* = 0.800), suggesting that the sample was representative of the population from which it was drawn [30]. In all, 80 women dropped out of the study prior to completing the final follow-up interview at 26 weeks however, there were no differences in the age, level of education and chosen feeding method of those who completed or withdrew from the study (data not reported). Data for the duration of any and full breastfeeding for women who withdrew were censored in accordance with the woman's status at the time of last contact, allowing all participants to be included in the survival analysis.

Almost all women (92.5%) initiated breastfeeding, and at six months, just over one third of all infants (39%) were receiving some breast milk and only 2% of infants had been fully breastfed to 26 weeks (Table 1). As not all women had ceased breastfeeding it was not possible to estimate mean duration of breastfeeding, however the median duration of any breastfeeding was 13.9 weeks and the mean duration of those women who had stopped breastfeeding before 26 weeks was six weeks. The median duration of full breastfeeding was less than one week with 50% of infants having received formula feeds within the first week post-partum.


**Table 1.** Prevalence a of full and any breastfeeding at selected ages.

a Survival analysis using censored cases.

The majority of women who initiated breastfeeding (*n* = 345) were between 25 and 34 years of age (64.3%), born in Kuwait (54.2%) and had completed 12 or more years of education (78%). Just over one third of women had delivered by Caesarean section (36.5%). Delayed initiation of breastfeeding and prelacteal feeding were the norm amongst this cohort with just over one half of women (52.8%) initiating breastfeeding more than 24 h after giving birth and the majority of infants (88.7%) receiving formula as either their first feed and/or at some time during the first three days of their hospital stay (Table 2). Only one third of women who initiated breastfeeding left hospital fully breastfeeding their infant, with the majority of women (59.4%) partially breastfeeding. A small number of women (*n* = 29, 8.4%) who initiated breastfeeding left hospital exclusively formula feeding.
