The present study was a prospective longitudinal cohort study of women from one to 4 months after childbirth. Arab and Jewish women residing in towns or villages in the northern part of Israel were recruited for interviews 1 month after childbirth at the Ministry of Health’s (MOH) Mother and Child Health Clinics (MCHC), which 95 % of new mothers visit. The interviews were performed from June 2015 to January 2016.
Sampling was performed in two stages. First, towns and villages, both Arab and Jewish, were chosen randomly from the two to four socioeconomic status (SES) ranking of towns (a 10-rung scale where 1 is the lowest and 10 the highest) calculated by the Israel Bureau of Statistics for the year 2013 . Most Arab towns are within this range of SES, therefore in order to match the two groups, a random selection was taken from Jewish towns and villages of the same SES rank. We chose one large Arab town (Nazareth), one large Jewish town (Afula), four Arab villages and three Jewish villages. All MCHCs in the chosen towns and villages were sampled. The second level of sampling included women visiting the MCHCs with their one-month-old baby. Nurses randomly recruited women answering the following inclusion criteria: mother, baby and all other children were healthy, the baby was born 1 month before and was delivered after the 37th week of pregnancy, the mother was living with a spouse, and gave informed consent to both the initial and follow-up interviews. Using a predetermined script, a research assistant explained the study and asked the women to give written informed consent. The women interviewed agreed to answer a second survey 3 months later. Those who agreed to answer the questionnaire were given it to fill out by a research assistant while waiting for their appointment with the nurse. Three months later they were contacted by phone and interviewed. The women had provided their telephone numbers for the follow-up interview on a separate form to maintain participant anonymity and a four-number identification code was added to the questionnaire to enable the coupling of the two interviews. The first interview lasted about 15–20 min and the second 8–10 min.
Research assistants were trained by one of the lead researchers who is proficient in both qualitative and quantitative research methods. As the first interview was self-administered, the assistant only answered questions if something was not understood. For the phone interview, research assistants were told to keep to the exact wording of the questionnaire, not help the women chose the answer, and to keep the same tone of voice when reading all possible responses. Before both interviews, the women were told that there is no correct or incorrect answer, that all answers are acceptable, and that their answers would be kept completely anonymous.
For the first interview, the response rate was 90%—92.6% for Arab women and 87.6% for Jewish women. Altogether 501 women were approached and 450 were interviewed (225 Jewish women and 225 Arab women). When approached 3 months later, 9.97% (23 Arab women and 22 Jewish women) of those interviewed did not answer the follow-up survey. Therefore, the response rate for the second interview was also 90%. This left 203 Arab women and 202 Jewish women in the follow-up sample. The non-respondent women were not significantly different from those that agreed to answer the questionnaire 4 months after childbirth regarding age, education, income, work status, number of children and obstetric characteristics.
The questionnaire included 9 parts: socioeconomic and demographic status, obstetric characteristics, social networks and support, negative social interactions, perceptions of customs and traditions intended to help the mother cope after childbirth, marital satisfaction, sense of parental competence, breastfeeding and health outcomes. The questionnaire was developed based on previous questionnaires and qualitative interviews with mothers from these population groups . The original Hebrew questionnaire was translated into Arabic and back-translated into Hebrew to assure the correct translation. A pretest was performed for both the Hebrew and Arabic versions of the questionnaire. Very small changes were made to the two versions as a result of the pretest. The second interview utilized a shorter version of the questionnaire with a focus on health outcomes.
Socioeconomic and demographic status
Included age, education, ethnic group (Arab or Jewish), years of marriage, work status (work out of home or not), income (above average, average or below average—compared to Israel’s net mean household income of 14,000 New Israeli Shekels), and place of residence (town or village).
Described if the pregnancy was achieved naturally or with medical help, if the delivery was normal or assisted (vacuum, forceps, elective or emergency caesarean section), if the mother fully or partially breastfed 1 month after childbirth, and number of children.
A social support scale was created by combining 11 items from Sherbourne & Stewart (1991) that were adapted to adequately measure social support after childbirth, and four items that were developed using findings from the initial qualitative study. The scale included three items depicting emotional support and eight items depicting instrumental support (See supplementary file 1).
The four additional items measured instrumental social support and inquired about having someone to help the mother take care of her older children, help take care of the baby, help so the mother could sleep and teach the mother how to take care of the baby and herself.
The available answers were on a range of five, from ‘never’ (1) to ‘always’ (5). Cronbach’s alpha was 0.84, 0.82 and 0.89 for instrumental, emotional and general social support (mean of both types of support) among Jewish women respectively, and 0.92, 0.85, and 0.92, respectively among Arab women.
The 19-item negative interaction questionnaire was based on Brooks and Dunkel’s 2011 model of social negativity . The questionnaire used a five-point Likert scale and included three dimensions: conflicts (8 items), insensitivity (7 items), and interference (4 items). The items were based on preliminary qualitative interviews and two previous studies on social support [24, 25, 22]. Principal component factor analysis was performed. Three components emerged fitting the theoretical model. Cronbach’s alpha was calculated separately for each component and for the entire scale. Among Jewish women, Cronbach’s alpha values were 0.95, 0.94, 0.83, and 0.96 for conflicts, insensitivity, interference, and the total scale, respectively. For Arab women, they were 0.94, 0.94, 0.74 and 0.96 respectively. Principal component analysis was performed with Varimax rotation. The analysis fitted the theoretical sub groups in both languages (Hebrew and Arabic) explaining 67.4 and 64% of variance respectively. (See supplementary file 1).
Sources of support and interactions
The women were asked to rate the levels of support and negative interactions they experienced from the various people around them—spouse, family, the spouses’ family, friends, neighbors and acquaintances—on a five-level scale from ‘does not help at all’ (1) to ‘is very helpful’ (5). (See supplementary file 1).
Perceptions of customs
This variable described how the women were impacted by postpartum customs and was measured using 19 items developed from the qualitative interviews. Women had to rate their agreement with the items (i.e. “To what extent do the customs and traditions that are performed for postpartum women enable you to rest” and “…cause you to be stressed”) on a scale of five from ‘very much’ (1) to ‘not at all’ (5). Four items were in the opposite direction and were reversed for analysis. Cronbach’s alpha was 0.84 among Jewish women and 0.90 among Arab women. (See supplementary file 1).
Self-reported health (SRH)
The women were asked to rate their current general health on a four-level scale, from very good to not good. For the logistic regression analysis, very good health was coded as 1 and the other three levels were coded as 0, as a high percentage of women (76% of Jewish women and 67% of Arab women) rated their health as very good and only a small percent as less than very good.
The women were presented with a list of 12 problems and asked to rate if they suffered from each problem, on a five-level scale, from ‘not at all’ (1) to ‘very much’ (5). The problems included: muscle pain, stomach pain, lower back pain, neck and shoulder pain, problems sleeping, pain in genitalia, constipation, emotional exhaustion, and feeling worried, cross and stressed . A mean score was calculated 4 months after childbirth with a Cronbach’s alpha of 0.91 for Jewish women and 0.83 for Arab women. (See supplementary file 1).
Analyses were performed with SPSS V21. T-test, chi2 tests, Man-Whitney and Wilcoxon tests were used to measure differences between Arab and Jewish women and between the first month and fourth month after childbirth. Spearman’s correlations were used to examine the relationship between health outcomes and social support, negative interactions and perceptions of customs and traditions. Multivariate linear regression models were run to assess the variables predicting health problems 4 months after childbirth and logistic regression models were run to assess factors predicting SRH 4 months after childbirth.