Demographic
Factors
Influencing
the Initiation
of Breast-Feeding
in an Israeli
Urban
Population
Emanuel
Birenbaum,
MD, Camil
Fuchs,
PhD,
and
Brian
Reichman,
MB, ChB
From the Department of Neonatology, The Chaim Sheba Medical Center, Tel Hashomer, and the Sackler School of Medicine and the Department of Statistics, School of Mathematical Sciences, Tel Aviv University, TelAviv, Israel
ABSTRACT. The influence of demographic and prenatal
factors on the initiation of breast-feeding in an urban
Israeli population was evaluated by interviewing 1,000
parturients prior to discharge from the hospital. The
mothers were all Jewish, none were single, and none were social welfare dependents. Of these mothers, 72% were
breast-feeding, 6% stated an intent to breast-feed, and
22% were formula-feeding their infants. A significantly
increased rate of breast-feeding was found among
moth-ers with the following characteristics: those of orthodox
religious belief, high educational level, in the academic
and paraacademic professions, nonsmokers, those who
worked outside of the home during the pregnancy, those who had previous breast-feeding success, and mothers
whose husbands’ attitude toward breast-feeding was
pos-itive. The decision to breast-feed was made prior to
delivery in 85% of mothers. Participation of the mother
and/or father in antenatal preparation courses did not
significantly influence the initiation of breast-feeding.
Multivariant analysis with a stepwise logistic regression
model delineated the four factors most significantly
as-sociated with the initiation of breast-feeding’ positive
spousal attitude toward breast-feeding, orthodox religious
belief, nonsmoking, and work outside of the home during the pregnancy. The expected probability for initiating
breast-feeding was computed for the various
combina-tions of these four categories and ranged from .94 with
all factors present to .33 in the absence of these
charac-teristics. Pediatrics 1989; 83:519-523; breast-feeding, Is-rael.
Programs designed to promote breast-feeding
practice have been suggested by national associa-tions of pediatricians.’3 Success of these programs
Received for publication Jan 18, 1988; accepted March 29, 1988. Reprint requests to (E.B.) Department of Neonatology, Chaim Sheba Medical Center, Tel Hashomer 52621, Israel.
PEDIATRICS (ISSN 0031 4005). Copyright © 1989 by the American Academy of Pediatrics.
will depend on the characteristics of each commu-nity. Analysis of the relationship between specific population characteristics and breast-feeding atti-tudes will help to define groups of mothers with a
high probability of breast-feeding and those likely
to be not breast-feeding their infants. The purpose of this study was to evaluate the demographic and prenatal factors influencing the initiation of breast-feeding in an urban Israeli population.
METHODS
The Chaim Sheba Medical Center is the regional medical center for the eastern greater Tel Aviv area, with approximately 5,000 deliveries a year. Mothers are generally selected for delivery at the hospital through a regional data base that controls patient flow to the various maternity hospitals. This selec-tion is based on information received from local antenatal clinics and is determined predominantly by the place of residence ofthe subjects. In addition, approximately 30% of patients received delivery cards through the high-risk pregnancy follow-up clinics of the hospital.
This study consisted of interviewing 1,000
par-turients prior to discharge from the hospital. Only
mothers who delivered healthy infants weighing more than 2,000 g were included. A prepared ques-tionnaire was administered to the mothers by trained nonmedical volunteers. Twice a week, on
predetermined days, all mothers whose discharge
was planned for the following day were interviewed. The questionnaire included demographic informa-tion such as level of education and employment, personal data such as age, parity, religious beliefs, smoking habits, and profession, and information
was explained to the mothers prior to the interview and no subjects refused to participate.
The relationship between breast-feeding and pos-sible predicting factors was examined. The response variable (breast-feeding) was categorized into three levels of response: breast-feeding, not breast-feed-ing, and intent to breast-feed. In a primary analysis, the relationship between the breast-feeding re-sponse and individual predictors (demographic,
personal, and antenatal) was evaluated by
x2
analy-sis. In a second step, the interdependence of the categorical variables was studied by multidimen-sional contingency table analysis. Preliminary
analysis showed that there were few observations
in the “intend to breast-feed” group (n = 68). We
thus decided to reduce the dimension of the table
and refer to the respondents’ two groups: breast-feeding and not breast-feeding. Based on the results of the primary analysis, the following eight van-ables were selected for creation of the multidimen-sional contingency table: breast-feeding (yes, no), religious belief (orthodox, traditional, secular), hus-band’s attitude toward breast-feeding (favorable, against, indifferent), smoking habits (yes, no, pre-vious smoker), number of children (one, two, three or more), education (primary, high school, college), age (less than 30 years or more), and working during pregnancy (yes, no).
The stepwise procedure used for model fitting in this study included a strategy for deciding on the redundancy of a predictor and on the reduction of the dimension of the table to be analyzed. Following the screening stage, Goodness of Fit tests were used to select the appropriate model. The model fitting
strategy was as follows: first stepwise logistic
regressions were fitted to the entire table to detect the most significant predictors. A series of tables were then created with those predictors, the re-sponse variable, and one of the remaining predic-tons. In this manner, we studied the contribution of each of the extra variables to the explanation of the variance of the response in condensed tables. The
computed program BMDPLR4 was used to fit the
stepwise logistic regression and the BMDP4F5 was used for the log linear models.
RESULTS
All of the interviewed mothers were Jewish and 31% defined themselves as orthodox religious (Ta-ble 1). Almost all mothers had completed high school and 52% had attended a college or higher level educational institution. Only 15% of the sub-jects did not have a defined occupation outside of the home and 67% of the mothers were employed outside of their homes during the present
preg-nancy.
At the time of the interview, 729 mothers were breast-feeding their infants, 68 intended to breast-feed, and 203 mothers were solely formula-feeding their infant. Most ofthe mothers (85%) had decided how to feed their infants before the delivery. The influence of individual demographic and antenatal factors on the percentage of mothers breast-feeding prior to discharge from the hospital is shown in Table 1. A higher proportion of breast-feeding was found among 20- to 30-year-old mothers with col-lege education, and mothers in an academic or paraacademic profession. A significantly higher proportion of mothers who worked outside of the home during pregnancy breast-fed their infants; however, plans to return to work after the delivery did not influence the initiation of breast-feeding. A higher incidence ofbneast-feeding was found among orthodox religious mothers. Mothers who smoked and previous smokers breast-fed less than non-smokers.
Approximately half of the mothers and 25% of fathers attended antenatal preparation courses.
Participation of either mothers or fathers in these courses did not influence the initiation of
breast-feeding. However, mothers who stated that their husbands’ attitude toward breast-feeding was posi-tive had a significantly higher rate of breast-feeding than those whose the husbands’ attitudes were neg-ative or indifferent. Successful breast-feeding of previous children had a strong positive influence on the initiation of breast-feeding in the present child.
In the stepwise regression analysis, the husbands’ attitude was the first variable to be entered in the model. The decrease in the
x2
value due to this variable was 42.5 (Table 2). Husbands’ opinion had the highest association with the response variable. In the following steps, the variables religious belief and smoking were included in the model withx2
values of 22.4 and 12.8, respectively. The
x2
values for the remaining predictors, age, educational level, number of children, and work during pregnancy, were all less than 8.00, which, although statistically significant, indicated a lower association with breast-feeding. In the second stage of the analysis,we created the five-way tables with the response variable breast-feeding, the three most significant predictors (husband’s attitude, religious belief, smoking), and one of the four remaining variables. Although the inclusion of age, educational level, or
number of children was of borderline significance, the addition of work was highly significant. Fur-thermore, the model that included the predictors,
husbands’ attitude, religious belief, smoking, and
TABLE 1. Demographic Factors and Rates of Breast-Feeding in the Hospital
Variable No. (%) of % of Responders P Value
Responders* Breast-Feeding Maternal age (yr)
18-20 44 (4.4) 63.6
21-25 274 (27.4) 75.9 <.05
26-30 342 (34.2) 75.6
>30 340 (34.0) 68.8
Ethnic origin
Oriental 417 (44.6) 72.2
Occidental 493 (55.4) 72.8 NS
Parity
1 344 (34.9) 75.5
2 263 (26.6) 67.2
3 197 (19.9) 71.6
4 75 (7.6) 77.3
5 or more 109 (11.0) 78.9 <.005
Education
Elementary 38 (3.8) 68.4
High school 444 (44.4) 67.3 <.005
Higher level institution 518 (51.8) 77.9
Religion
Secular 412 (41.4) 66.5
Traditional 273 (27.5) 70.3 <.001
Orthodox 310 (31.1) 83.9
Smokes
Yes 138 (14.0) 59.3
Formerly 89 (9.0) 66.7 <.001
No 758 (77.0) 75.9
Profession
Hosuewife 153 (15.4) 66.0
Office workers/artisan 440 (44.2) 71.4 <.05
Academic/para-academic 402 (40.4) 77.6
Work outside home during pregnancy
No 315 (32.3) 67.9
Yes 661 (67.7) 75.9 <.02
Plan to return to workt
No 125 (17.2) 72.0
Yes 606 (82.8) 75.1 NS
Antenatal preparation course
No 452 (45.6) 71.7
Yes 302 (30.5) 76.2 NS
With previous pregnancy 237 (23.9) 70.5
Husband participated in
anten-atal course$
No 246 (46.4) 73.6
Yes 284 (53.6) 72.0 NS
Husband’s attitude toward breast-feeding
Negative 19 (1.9) 47.4
Indifferent 221 (22.2) 56.6 <.001
Positive 757 (75.9) 78.6
Previous breast-feeding success
None 130 (20.3) 36.9
Partial 95 (14.8) 64.2 <.001
All 416 (64.9) 84.1
* Not all subjects responded to every question.
1 Includes only subjects working outside of their homes.
:1:Includes only husbands whose wives participated in an antenatal course.
for each combination of categories of these four pregnancy, and whose husbands’ attitudes were
predictors is shown in the Figure. For example, the positive was .94. The estimated probability for estimated probability ofbreast-feeding for orthodox women in the category with these characteristics
df P Value for the
Added variable
P Value for the Model (Including Variable Added) Most significant
Husband’s attitude 42.5 1 .000 Religious belief 22.4 2 .000 Smoking habit 12.8 2 .002 Additional
Working during pregnancy 7.87 1 .005 .535
Parity 7.68 2 .022 .079
Mother’s age 5.26 1 .022 .169
Educational level 4.92 1 .026 .060
W-WORK
NW-NO WORK *-NO SUBJECT
Figure. Estimated probabilities of breast-feeding for
combination of major predicting factors.
TABLE 2. Stepwise regression analysis
x2
Variable
DISCUSSION
Current research in developed countries such as Finland,6 Sweden,7 and Norway8 showed that more than 90% of mothers initiate breast-feeding in the
hospital. In countries with heterogeneous popula-tions such as the United States, an increasing trend
toward breast-feeding has been notedY’3 A wide variability of the proportion of breast-feeding
mothers has been noted, however, ranging from
36.8% to 71.8% for lower and higher socioeconomic
groups, respectively.” The results obtained in this
study were similar to those reported in Central
European countries’4 and among mothers with a
high educational level in the United States.”
Al-though our population was heterogeneous, includ-ing a wide range of socioeconomic and ethnic groups, several common characteristics unified this group. A high percentage of mothers were high
school graduates, none were social welfare
depend-ents, and all were married. Furthermore, free
an-tenatal clinics were attended by more than 99% of
our patients, and all mothers received postnatal maternity benefits including 3 months of fully paid maternity leave.
Of the various demographic factors evaluated, we were able to define four major factors influencing
the initiation of breast-feeding: religious belief,
smoking habits, working during pregnancy, and husbands’ attitude toward breast-feeding. Combi-nation of these factors resulted in an estimated probability for breast-feeding ranging from .33 to .94. The high proportion of either orthodox or tra-ditional religious mothers is a characteristic of the population delivering in our hospital. Compared
with either traditional or secular mothers, a
signif-icantly higher proportion of orthodox mothers breast-fed their infants. The positive association between religious belief and breast-feeding has been
reported in developing 516 We believe
that the religious mothers among our patients are
powerfully influenced by the traditional acceptance
of breast-feeding as the obvious mode of feeding and by the established and written priority of breast-feeding over other family events such as a further pregnancy or remarriage.
The negative influence of smoking has previously
been Although only a small
pro-portion of our patients smoked, this negative
influ-ence was noted in the various demographic groups studied. Only a small number ofthe orthodox
moth-ers smoked and thus the probability for
breast-feeding in religious mothers who smoked could not be calculated for all of the groups shown in the
Figure. The relationship of maternal employment and breast-feeding has been evaluated in a number of studies. Martinez and Nalezienski in 1981 found a lesser rate of breast-feeding among mothers who worked outside of the home. More recent surveys
show that maternal employment is associated with
a higher rate of breast-feeding, 11 17,18
Employment per se does not appear to be a reason
the mother’s breast-feeding experience.’9 All work-ing mothers in Israel receive 3 months’ fully paid maternity leave with a legal option of a further 9
months’ unpaid leave without compromising their
employment status. This security may have a pos-itive influence on the working mother’s decision to breast-feed.
Husbands’ opinions regarding breast-feeding ap-peared to be the major determinant for the initia-tion of breast-feeding in our population. Fathers were not specifically interviewed and the answers represent the mothers’ opinions of their husbands’ attitudes. Obviously, their answers may be biased by the mothers own decision. Nevertheless, in 55% of the nonbreast-feeding mothers, the husbands’ attitude was reported as being positive. It appeared that the fathers support and attitude toward breast-feeding was likely to have a positive influence. This fact has previously been confirmed2#{176} and the Amer-ican Academy of Pediatrics recommended educa-tion of young boys about breast-feeding.2
The influence of maternal “ number of children,9”#{176}”7’20’2’ and maternal educational
10 14,20-22 on initiation of breast-feeding were
reported. It is important to emphasize that the model selected in our study does not suggest that the predictors age, education, and number of chil-dren were not related to breast-feeding, but rather that, once the other independent variables were included in the model, the extra contributions of these predictors was marginal.
The population surveyed in this study was unique in its characteristics and the factors that influence the initiation of breast-feeding will in all likelihood differ in other settings. However, in other
popula-tions, this model may enable definition of the
spe-cific demographic characteristics of mothers who are likely to breast-feed and, conversely, those who are likely not to initiate breast-feeding. Early iden-.tification of these mothers during pregnancy and
in the immediate postpartum period may enable an active intervention program to promote breast-feeding practices.
REFERENCES
1. American Academy of Pediatrics, Committee on Nutrition: Encouraging breast-feeding. Pediatrics 1980;65:657-658
2. American Academy of Pediatrics, Policy Statement Based on Task Force Report: The promotion of breast-feeding.
Pediatrics 1982;69:654-661
3. Canadian Paediatric Society, Nutrition Committee and
American Academy of Pediatrics, Committee on Nutrition: Breast-feeding. Pediatrics 1978;62:591-601
4. Dixon W: Stepwise regression, in BMDP, statistical
soft-ware. Berkeley, CA, University of California, 1983, pp
251-263
5. Dixon W: Frequency tables. in BMDP, statistical software. Berkeley CA, University of California, 1983, pp 143-206 6. Stahlberg MR: Breast-feeding and social factors. Acta
Pae-dkztr Scand 1985;74:36-39
7. Sj#{246}lin5, Hofvander Y, Hillerwik C: Factors related to early termination of breast-feeding. Acta Paediatr Scand 1977; 66:505-511
8. Meberg A, Wilgruff 5, Sande HA: High potential for breast-feeding among mothers giving birth to pre-term infants.
Acta Paediatr Scand 1982;71:661-662
9. Martinez GA, Nalezienski JP: 1980 uptake: The recent trend in breast-feeding. Pediatrics 1981;67:260-263
10. Martinez GA, Dodd DA, Samartgedes JA: Milk-feeding
pat-terns in the United States during the first 12 months of life.
Pediatrics 1981;68:863-868
11. Martinez GA, Krieger FW: 1984 milk-feeding patterns in
the United States. Pediatrics 1985;76:1004-1008
12. Hendershot GE: Trends in breast-feeding. Pediatrics
1984;74(suppl):591-614
13. Saret HP, Bain KR, O’LearyJC: Decisions on breast-feeding
or formula-feeding and trends in infant-feeding practices.
Am J Dis Child 1983; 137:719-725
14. Florack E, Obermann-De Beer G, van Kampen-Donker M, et al: Breast-feeding, bottle-feeding and related factors. Acts
Paediatr Scand 1984;73:789-795
15. Forman MR: Review of research on the factors associated with choice and duration of infant-feeding in less developed countries. Pediatrics 1984;74(suppl):667-694
16. Pursall EW, Jepson ME, Smith BAM, et al: Breast-feeding
and mother’s education. Lancet 1978;30:734-735
17. Simopoulus AP, Grave GD: Factors associated with the choice and duration of infant-feeding practice. Pediatrics
1984;74(suppl):603-614
18. Auerbach KG, Guss E: Maternal employment and
breast-feeding. Am J Dis Child 1984;138:958-960
19. Katcher AL, Lanese MG: Breast-feeding by employed moth-ers: A reasonable accommodation in the work place.
Pedi-atrics 1985;75:644-647
20. Easthain E, Smith D, Poole D, et al: Further decline of breast-feeding. Br Med J 1976;1:305-307
21. Sauls HS: Potential effect of demographic and other van-ables in studies comparing morbidity of breast-fed and
hot-tle-fed infants. Pediatrics 1979;64:523-527
22. Notzon F: Trend in infant-feeding in developing countries.