This population-based case-control study was conducted on 2463 mothers, including 668 cases and 1795 controls, referred to a health care center in several provinces of Iran, namely, Fars, Hormozgan, Kermanshah, Hamadan, Kohgiloyeh, and Boyerahmad, Yazd, Southern Khorasan, Golestan, and city of Mashhad (Fig. 1). A rural health care center is a health facility in a village that provides health care for approximately 9000 people of that village and several neighboring villages. Health care providers at a rural health care center include a general physician and public health and midwifery experts. The rural health care center supervises and supports health care facilities in villages and is linked to its superior urban health care center. An urban health care center is a health facility in cities providing care to approximately 12,500 people. Health care providers, including a general physician and public health and midwifery experts, provide laboratory, pharmaceutical, radiological, and medical care in the urban health care centers. Experts at rural and urban health care centers register the provided health care to every family in their health records, such as health care for pregnant women. However, the registered information in the family’s health records were insufficient, and hence we collected additional data through interviews with the study participants.
The case group was defined as women who had preterm birth in a recent pregnancy, and the control group was defined as women who had full-term birth in a recent pregnancy [13,14,15]. The sample size ratio in the control and case groups was 3:1. Data were collected through interviews according to a check list containing demographic information (mother’s age, ethnicity, occupation and level of education, place of residence, and consanguineous marriage) and information on the previous pregnancies (the outcome of previous pregnancy, mode of delivery, and interpregnancy interval).
Study subjects were recruited through a multistage cluster sampling method. In the first stage regarding geographical divisions of Iran, nine clusters (provinces) were randomly selected. In the second stage, in each of the nine clusters (provinces), four clusters (cities) were randomly selected from the north, south, east, west, and central areas. In each city, two health care centers (one urban and one rural health care center) were randomly selected. In each health care center, 10 check lists were filled in by well-trained interviewers according to a protocol. In each center, data collection process was conducted simultaneously on the same day for cases and controls. Data of the control group were collected from a random sample of mothers referring to the health care center. If < 10 cases were available in each health care center, the remaining check lists were filled in the nearest center, and if there were > 10 cases, the check lists were filled in for a random sample of mothers. We tried to maintain the same size for the case and control groups.
According to literature review, considering mother’s age > 35 years as a risk factor (p0 = 0.3, p1 = 0.44, z0.95 = 2, z(1-β) = 0.8, design effect = 2) [16] and using the proportion determination formula, the sample size was estimated as 370 for each study group. In this study, the association between preterm birth and 14 independent variables was evaluated. Therefore, taking into account an additional 20 samples for each independent variable, the total sample was calculated as 650 for each study group. The sample size was sufficient considering > 80% power of study.
Sociodemographic variables
Sociodemographic information included mother’s age (age < 35 or ≥ 35 years), place of residence (urban vs. rural area), occupation (housewife/employee/cowhand/farmer/carpet weaver), level of education (illiterate/primary school/intermediate school/high school/academic gradation), ethnicity (Turk/Lor/Arab/Balooch/Torkaman/Fars/Kurd or others), and marriage (consanguineous vs. nonconsanguineous).
Information on the previous pregnancies
This included history of abortion, stillbirth or cesarean section (yes/no), interpregnancy interval (the first pregnancy/< 1 year/1–3 years/> 3 years), BMI (normal/low weight/overweight/obese: grade 1, grade 2, and higher), and cycles of menstruation period (regular/irregular).
The outcome variable
Preterm birth was the outcome variable, which was ascertained through questioning the exact gestational age at the time of birth.
Statistical analysis
Descriptive statistical tests were performed for socio-demographic and pregnancy-related variables. Bivariate analysis was performed to identify the association of dependent and independent variables. Odds ratio was computed to see the strength of association between preterm birth and each of categorical variables. Adjusted odds ratio and their 95% confident interval were calculated by including all exposures with p value < 0.3 in the multivariate model to control for confounding effects [17]. Data were analyzed using SPSS version 19, with two-tailed tests at p ≤ 0.05 level of significance.