Breast-feeding is considered internationally as the ideal method of feeding for infants. The WHO(1) and the American Academy of Pediatrics (AAP)(2) recommend that infants are exclusively breast-fed for the first 4 or 6 months of life. Exclusive breast-feeding, as defined by WHO, is ‘no other food or drink, not even water, except breast milk for at least 4 and if possible 6 months of life, but allows the infant to receive drops and syrups (vitamins, minerals and medicines)’. After the suggested period, mothers can gradually introduce liquids and solids to complement breast-feeding(1). The AAP advises mothers to continue breast-feeding for at least 1 year(2), while WHO emphasizes the importance of stretching the duration to the first 2 years of life(1).
It is a well established fact that breast-feeding has priceless advantages for the growth and development of the baby, the health of the mother and the well-being of society(2). Despite these essential benefits, breast-feeding rates are low in various regions of the world. According to the WHO global databank on breast-feeding, only 35 % of infants in ninety-four countries are exclusively breast-fed under the age of 4 months(3). To improve breast-feeding rates, a number of international studies have invested in determining the predictors of successful breast-feeding practices. Such investigations are essential in identifying target groups and modifiable risk factors for effective intervention programmes, such as counselling and educational programmes that have proved to be successful(Reference Haque, Hussain, Sarkar, Hoque, Ara and Sultana4–Reference Gross, Caulfield, Bentley, Bronner, Kessler, Jensen and Paige7).
Literature on breast-feeding presents several predictors related to maternal characteristics and psychological state, infant characteristics, cultural customs, sociodemographic status and medical-related factors(Reference Dubois and Girard8–Reference Ford and Labbok16). Although physician support for breast-feeding has been shown to improve breast-feeding rates(Reference Freed, Clark, Lohr and Sorenson17), very few studies have explored physician-related factors as potential breast-feeding determinants. Sex of the paediatrician, for instance, has not been previously researched as a possible predictor, despite the fact that it has been found to be an influential factor for different health outcomes(Reference Lurie, Slater, McGovern, Ekstrum, Quam and Margolis18, Reference Maheux, Haley, Rivard and Gervais19).
In Lebanon, data on the prevalence and predictors of breast-feeding remain scarce. Thus the present study aimed to assess the prevalence and predictors of breast-feeding at 1 and 4 months of infant age while exploring the potential role of the sex of the paediatrician, using data from a prospective cohort project conducted by the National Collaborative Neonatal and Perinatal Network (NCPNN) in Beirut, Lebanon.
Experimental methods
Study design and population
The present study used data from the First Year of Life Follow-Up Study. It involved the participation of a close birth cohort of 1320 healthy infants, aged 0–2 months, who were recruited over a period of 7 months (1 August 2001 to 29 February 2002) and followed-up during their first year of life. Infants were enrolled through the clinics and dispensaries of 117 paediatricians located in Beirut and its suburbs (known as the Greater Beirut area) and serving communities of different socio-economic status. The inclusion criteria of the First Year of Life Follow-Up Study were: (i) Lebanese nationality; (ii) single gestation; (iii) birth weight more than 2200 g; (iv) gestational age greater than 35 weeks; (v) absence of major congenital anomalies at birth; (vi) cared for in the hospital of birth and discharged alive within 5 days; (vii) age at first paediatrician visit between 0 and 2 months; and (viii) informed consent from the family.
Data collection
The cohort was followed using a total of seven questionnaires that were completed for each infant. These were: (i) a recruitment form containing basic information about the newborn infant and his/her mother at delivery; (ii) five follow-up questionnaires completed by the paediatricians at five different age intervals corresponding to routine visits at 0–2, 3–4, 5–7, 8–10 and 11–13 months, these included information on anthropometric measurements, vaccination, health-care visits, infectious diseases, injuries, breast-feeding, dental health and others; and (iii) a parental follow-up questionnaire, obtained by telephone call performed by a research assistant six months after the date of birth of the infant, which included information on parental sociodemographic characteristics, maternal postpartum complications and maternal lifestyle and obstetric characteristics.
In the event that an enrolled baby was lost to its paediatrician, research assistants would track the baby to other paediatricians and/or obtained the needed information from parents through follow-up telephone calls. A baby was considered lost to follow-up if he/she could not be traced through a paediatrician or by telephone.
Analysis
Information on breast-feeding was obtained at each of the five follow-up visits through a question on the current method of milk feeding. The breast-feeding definition adopted in this study was ‘full breast-feeding’ whereby the infant receives only breast milk without the supplementation of any non-human milk. No consideration about the intake of solid foods was made in our definition.
Based on the literature, the independent variables comprised the following: (i) sociodemographic predictors: religion, parental education and maternal working status; (ii) maternal characteristics: maternal age, parity, smoking during pregnancy, maternal height, pre-pregnancy weight, pregnancy weight gain and weight 6 months after delivery; (iii) infant characteristics: sex of the baby, gestational age and birth weight; and (iv) medical/delivery-related factors: admission status, mode of delivery, length of hospital stay, hospital of delivery and the sex of the paediatrician whom the study participants sought for medical care.
At the bivariate level, differences in the distribution of fully breast-fed babies were assessed among the different levels of each predictor. For continuous predictors, t tests with P < 0·05 were used, while cross-tabulations and odds ratios with 95 % confidence intervals were performed for categorical predictors. Logistic regression using stepwise modelling was computed on the predictors that proved significant at the bivariate level. An adjustment for dependencies within hospitals of delivery (clusters) was undertaken to account for the variations among them. Adjusted odds ratios and 95 % confidence intervals are reported for the final model. All analyses were computed in the Statistical Package for the Social Sciences statistical software package version 7.5 (SPSS Inc., Chicago, IL, USA) except for the cluster effect, which was performed using the STATA 6.0 statistical software package (StataCorp LP, College Station, TX, USA).
Ethical considerations
The First Year of Life Follow-Up Study was approved by the university Institutional Review Board at the American University of Beirut. Participation was on voluntary basis. After the full disclosure of the project to the parents, an informed consent was obtained. Information on the participants remained confidential and all analyses performed retained participant anonymity.
Results
A total of 1320 infants were recruited. The 0–2 months age interval questionnaire was completed for all 1320 newborns, while the 3–4, 5–7, 8–10 and 11–13 months follow-up questionnaires were completed for 1171 (88·7 %), 1,127 (85·4 %), 932 (70·6 %) and 1059 (80·2 %) infants, respectively. Table 1 presents the prevalence of full breast-feeding at the different age intervals. At the end of the first month of life, the rate of infants receiving full breast-feeding was 56·3 %. The rate dropped by more than half (24·7 %) at 4 months and reached 18·8 % by 6 months of age. At 1 year of life, the intake of breast milk as the only source of milk was minimal (6·7 %). Data on full breast-feeding at 1 month were missing in 19·3 % of cases. Comparison of sociodemographic, maternal, infant and delivery-related characteristics between those having no information on full breast-feeding status and those having information was performed. No significant difference between the two groups was evident except for maternal age, gestational age and weight gain during pregnancy. Subjects missing information on full breast-feeding at 1 month were mostly of secondary education, higher gestational age and lower pregnancy weight gain. Nevertheless, the difference between the two groups among the different levels of the three variables had no clinical significance.
Tables 2 and 3 present the bivariate results for the categorical and continuous predictors of full breast-feeding, respectively. Significant predictors for full breast-feeding at 1 and 4 months of infant age were all of the sociodemographic indicators, maternal age, parity, maternal height, mode of delivery, early discharge and sex of the paediatrician. Pregnancy weight gain was significant only at 1 month.
*95 % CI not calculated (reference category).
Stepwise regression analysis was performed on all variables that proved significant at the bivariate level. Table 4 shows the final regression models for the predictors of breast-feeding at 1 and 4 months of age. Analysis for the clustering effect of the hospital of delivery was performed; yet no differences in the regression estimates were evident. At 1 month, parity, maternal age and early discharge were the most significant predictors of full breast-feeding. Mothers having one child and mothers having more than two children were respectively two and three times more likely be breast-feeding their newborn than nulliparous mothers. Mothers aged <25 years were 2·5 times more likely than mothers aged >35 years to be breast-feeding their infant at 1 month. Newborns discharged within 48 h of delivery were twice as likely to be breast-fed at 1 month compared with those discharged later than 48 h. Religion had a borderline significance with breast-feeding at 1 month. Although the paediatrician’s sex was retained in the final model at 1 month, the association was not statistically significant. In contrast, at 4 months of age, paediatrician’s sex was a significant predictor of breast-feeding whereby women seeking health care from female paediatricians rather than male paediatricians were 1·5 times more likely to remain breast-feeding until 4 months. Maternal employment status also emerged as a specific predictor for the model at 4 months. Non-working mothers were more than two times as likely to continue breast-feeding at 4 months compared with working mothers. Among the other covariates, parity, religion and early discharge remained significant at 4 months, while maternal age was no longer significant. However, religion was a stronger predictor of breast-feeding at 4 months, where the odds of breast-feeding among Muslim mothers were about twice those among Christian mothers.
*95 % CI not calculated (reference category).
Discussion
In the present study population, the full breast-feeding rate at 1 month of age was low (56·3 %). At 4 months it dropped dramatically and remained very low until the end of the first year of life. When addressing the predictors of full breast-feeding at the multivariate level, early discharge, high parity, being Muslim and young maternal age proved significant predictors at 1 month of age. At 4 months, Muslim and non-working mothers, high parity, early discharge and seeking care from female paediatricians were the significant predictors.
The literature on breast-feeding is ample; yet this is one of the rare studies in the Middle Eastern and Arab region that looks into the predictors of breast-feeding using longitudinal data. Previous studies are mainly of cross-sectional nature and focus on the prevalence of breast-feeding(Reference Bagenholm, Kristiansson and Nasher20–Reference Osman and el Sabban23) and its predictors at the bivariate level(Reference El Mougi, Mostafa, Osman and Ahmed24–Reference Shahraban, Abdulla, Bjorksten and Hofvander27). In Lebanon, a recent cross-sectional study revealed comparable prevalence rates to those in the present study (52·4 % at 1 month and 23·4 % at 4 months), whereas the duration of exclusive breast-feeding was longer among rural mothers and those who attained lower education(Reference Batal, Boulghourjian, Abdallah and Afifi9). Particular to our population, maternal age was significantly associated with breast-feeding. Evidence in the literature provides consistent results of a positive association between breast-feeding duration and maternal age(Reference Dubois and Girard8, Reference Dulon, Kersting and Schach10–Reference Scott and Binns14). Thus the present results are not in agreement with the literature as the odds of full breast-feeding at 1 month in our study increased with a decrease in maternal age. However, a study in Germany reported that the odds of breast-feeding for less than 4 months were 3·53 times higher among young mothers (≤25 years) than older women(Reference Dulon, Kersting and Schach10).
A positive dose–response relationship between parity and breast-feeding rates was present at both time intervals. Association between breast-feeding and parity is not consistent in the literature(Reference Lande, Andersen, Baerug, Trygg, Lund-Larsen, Veierød and Bjørneboe11, Reference Scott and Binns14). The relationship ranges from a negative association in some studies(Reference Abada, Trovato and Lalu15, Reference Ford and Labbok16) to a positive association in others(Reference Jakobsen, Sodemann, Molbak and Aaby28, Reference Piper and Parks29). In agreement with our study, the odds of exclusively breast-feeding at 4 months in Lande et al.’s(Reference Lande, Andersen, Baerug, Trygg, Lund-Larsen, Veierød and Bjørneboe11) study was two times higher among mothers with three or more children than among mothers with one child. Parity might contribute to higher breast-feeding rates because the mother’s breast-feeding experiences from previous pregnancies provide her with more knowledge and self-confidence.
With reference to early discharge, the odds of breast-feeding increased among women discharged within 48 h of delivery. Early discharge was a significant factor for both the initiation of breast-feeding at 1 month and its continuation until 4 months of age. Although a number of studies suggest the absence of an association(Reference Madden, Soumerai, Lieu, Mandl, Zhang and Ross-Degnan30–Reference Kvist, Persson and Lingman36), Bussolati et al.(Reference Bussolati, Gambini, Musetti, Braibanti and Capuano37) and Margolis and Schwartz(Reference Margolis and Schwartz38) reported results similar to the present study. This can be attributed to the absence of rooming-in systems in the hospitals, unsupportive practices of health-care staff and the unmonitored promotion of formula milk by sales representatives(Reference El Mougi, Mostafa, Osman and Ahmed24).
While religion had borderline significance at 1 month, it was a stronger predictor of breast-feeding at 4 months of age. Religion thus played a significant role in the continuation of breast-feeding until 4 months of age, with Muslim mothers twice as likely to breast-feed compared with Christian women. Similarly, a study noted that Muslim mothers in Kenya breast-feed for the longest durations(Reference Mott39). This result can be explained by Islam’s encouragement of breast-feeding(Reference Bagenholm, Kristiansson and Nasher20, Reference Sharief, Margolis and Townsend25): it is stated in the Holy Qur’an that ‘a mother shall breast-feed her child for two years’ (Holy Qur’an 31:14). It is worth mentioning that religiosity, the degree to which a person practices his/her religion, might add further insight to the current data.
Concerning maternal employment status, non-working mothers were significantly more likely to breast-feed until 4 months (OR = 2·38) compared with working mothers. Likewise, housewives in Egypt (1981) breast-fed for longer durations than working mothers(Reference El Mougi, Mostafa, Osman and Ahmed24). In Malaysia (1995) as well, higher odds of breast-feeding at 6 weeks were evident among non-working mothers (OR = 1·48)(Reference Chye, Zain, Lim and Lim40). Evidently, non-working mothers have more time to breast-feed their children than their employed counterparts. This is exacerbated by the short maternity leave in Lebanon (range from 40 days to 2 months). Working mothers usually try to terminate breast-feeding before the end of their maternity leave, especially with the lack of facilities allowing expressing of breast milk at the work place.
A novel finding of the present study was the positive effect of female paediatricians on breast-feeding continuation until 4 months of age. At 1 month of age, the sex of the paediatrician was not significant in the multivariate model after controlling for other variables. However, its significance was evident in the 4 month model, whereby babies visiting female paediatricians had higher breast-feeding rates. Therefore, the sex of paediatrician has no effect on the initiation of breast-feeding but plays an important role in the continuation of breast-feeding. Although no literature is available on the effect of the sex of the physician on breast-feeding, some studies have examined its effect on other outcomes. Similar to our study trend, Lurie et al.(Reference Lurie, Slater, McGovern, Ekstrum, Quam and Margolis18) showed that women consulting female physicians are more likely to have a Pap smear performed and mammography screening tests. They also reported getting more counselling on the use of condoms as a preventive method for sexually transmitted diseases and undesired pregnancies(Reference Maheux, Haley, Rivard and Gervais19). According to a recent review article(Reference Roter and Hall41), female doctors are described to ‘engage in more active partnership behaviors, positive talk, psychosocial counseling, psychosocial question asking, and emotionally focused talk’. Female physicians also might encourage and support breast-feeding by reflecting on their personal breast-feeding experiences.
The present study has several strengths. Being a prospective cohort study, it was not exposed to recall bias and information on breast-feeding status was available as of the first months of life. The study also accounted for a considerable number of covariates that serve as potential predictors for breast-feeding. The confounding effect of many predictors was thus well contained. The sample size was large (n 1320) and relatively representative of the Greater Beirut area, as the cohort was recruited from the clinics and dispensaries of 117 paediatricians serving communities of different socio-economic status and of different geographic areas in Beirut. The heterogeneity of the sample, as a result of proper recruitment, allows for generalizability of the results to the Greater Beirut area. Moreover, according to the Central Administration for Statistics of the Lebanese Republic, there were 7096 births in total in the Greater Beirut area during the recruitment period (1 August 2001 to 29 February 2002). Since the present study recruited 1320 babies, it was able to capture 18·6 % of the total births during that period(42).
A limitation of the study was the inability to incorporate solid food intake in the definition of breast-feeding as per the WHO description of ‘exclusive breastfeeding’. There was also no measure of breast-feeding status at the hospital of delivery. The breast-feeding status before discharge aids in assessing hospital practices and its effect on future breast-feeding status. Besides the sex of the paediatrician, information on the training, knowledge, duration or experience and qualification of paediatricians would have provided a clearer understanding of the paediatrician’s role on breast-feeding.
Conclusions
In Beirut, Lebanon, low breast-feeding rates at 1 and 4 months of infant age have been explained by early discharge, young maternal age and high parity. Breast-feeding continuation up to 4 months of age has been further explained by religion, maternal working status and sex of the paediatrician. Additional research is warranted to investigate the interactions between female physicians and lactating mothers in maintaining breast-feeding in other populations. Our results also constitute the basis for designing interventions targeting policy makers, health professionals, mothers and researchers.
Acknowledgements
The present work was supported by grants from the WHO, the Lebanese National Council for Scientific Research (LNCSR), the University Research Board (URB), the Medical Practice Plan (MPP) and the Chairman Research Fund of the Department of Pediatrics at the American University of Beirut Medical Center and the Lebanese Pediatric Society. None of the authors had any competing interests. The authors thank the 117 paediatricians and parents who participated in this prospective cohort study. The authors would also like to acknowledge the nurses in every setting and the research assistants who collected and entered the data. B.A.S. did the analysis and wrote the first draft of the manuscript, H.T. participated in the study design and supervised the analysis and write-up, G.M. participated in the analysis and final editing of the manuscript, M.K. provided statistical and methodological help, M.K. and R.A. revised and edited the paper, Y.N. participated in data collection and K.A.Y. is the Principal Investigator of the Follow-Up Project; he produced the original study design and revised the manuscript.