Study design, period, and setting
An institutional-based, unmatched case control study was conducted from December 1, 2022, to March 30, 2023, at different public hospitals in the North Wollo zone in northeastern Ethiopia. The North Wollo Zone has a total of six public hospitals: one comprehensive, one general, and four primary hospitals. The study was conducted at Woldia Comprehensive Specialized Hospital (WCSH), Lalibela General Hospital, and Mersa Primary Hospital. The abovementioned public hospitals are providing continuous maternal and child health services, including ANC and delivery services, around the area.
Study population and eligibility criteria
All women who visited the selected public hospitals for labor care, had a gestational age of more than 28 weeks, were diagnosed with PROM by clinicians and were admitted to the labor and maternity wards of each hospital during the study period were enrolled as cases. Moreover, all women who visited the selected hospitals for labor care, had a gestational age of more than 28 weeks, were not diagnosed with PROM, were confirmed by clinicians and were admitted to the labor and maternity wards of each hospital during the study period were included as controls.
Those mothers who were initially assigned to the control group but later switched to the case group and those who were seriously ill during the data collection period were excluded from the study. Additionally, women who underwent artificial membrane rupture, who experienced maternal deterioration requiring immediate delivery, who experienced intrauterine fetal death, or whose cases were due to trauma or accidents were excluded.
Sample size determinations
The sample size was determined using the StatCalc menu of Epi Info version 7.2.5 statistical software for an unmatched case control study. The following assumptions were used: a 95% confidence level, 80% power, a case-to-control ratio of 1:2, a percentage of exposure among controls (proportion of no ANC visit among women without PROM), and a percentage of exposure among cases (proportion of no ANC visit among women with PROM). The percentages of patients with no ANC visits (27.9%) and controls with no ANC follow-up (14.0%) were taken from a study conducted in southern Ethiopia (15). Based on the above assumptions, the sample size was 321 (107 cases and 214 controls), and after adding for a 10% nonresponse rate, the final sample size for the study was 353 (118 cases and 235 controls).
Sampling techniques and procedures
Among the 6 public hospitals found in the zone, three public hospitals (WCSH, Lalibela Referral Hospital and Mersa Primary Hospital) were selected by the lottery method. Then, the number of cases and controls were allocated proportionally based on the same period of previous year pregnant women attended in each hospital. Therefore, from a total of 1707 study participants from the last year at WCSH, we selected 193 participants (64 cases and 129 controls); from 938 pregnant women at Lalibela General Hospital, we selected 106 participants (35 cases and 71 controls); and from Mersa Primary Hospital, a total of 478 pregnant women were admitted last year, and we selected 54 participants (18 cases and 36 controls). Finally, all patients were selected until the sample size was reached. Moreover, controls were selected by systematic random sampling. Following the admission of a specific case to the unit, two controls were randomly selected using a lottery method from a list of controls admitted to the same unit, and the kth of those controls was selected accordingly. The process of control selection commenced promptly after the selection of each patient.
Operational definitions
Mid-upper arm circumference: The nutritional status of the mothers was evaluated by measuring the mid-upper arm circumference, which was assessed at the midpoint between the tips of the shoulder and the elbow of the left arm using standard MUAC tape. The measurements were taken to the nearest 0.1 cm and recorded in the provided questionnaire as less than 23 cm and greater than or equal to 23 cm (24).
Chronic cough: Participants with a persistent cough that lasts for at least eight weeks and often much longer (25).
Gestational hypertension was defined as a pregnant mother with a blood pressure greater than or equal to 140 mmHg and/or diastolic blood pressure greater than or equal to 90 mmHg on two separate occasions at least four hours apart after 20 weeks of pregnancy when the previous blood pressure was normal (26).
Suspected sepsis was defined as a temperature below 36.0°C or above 38.0°C for a minimum of 2 hours and 1 hour, respectively; white blood cell counts below 4 × 10^9/l or above 12 × 10^9/l, or more than 10% immature (band) forms; a heart rate surpassing 90 beats per minute for at least 1 hour; and a respiratory rate exceeding 20 breaths per minute for at least 1 hour. Sepsis was characterized by the presence of a minimum of two of these criteria in addition to the clinical suspicion of infection as determined by experienced senior clinicians (27).
Urinary tract infection was defined as a positive result for nitrite and leucocyte esterase on a urine dipstick and the presence of red and white blood cells, squamous cells and bacteria on a urine microscope (28).
Data collection methods and procedures
Interviewer-administered questionnaires were used for data collection. The questionnaires were initially prepared in English and then translated to Amharic and returned to English to check consistency. A review of the cards was also performed to obtain information on some variables. Three midwifery professionals at each hospital were recruited for data collection. One BSc midwife in each hospital was assigned as a supervisor to supervise the data collection throughout the process at each health institution.
Data quality control
To ensure the quality of the data, data collection tools were used after intensive review of the relevant literature. Pretesting of the questionnaire was carried out on 5% of the sample at the Haik Primary Hospital. Based on the pretest results, all relevant corrections were made. One-day training was given to the data collectors and supervisors about confidentiality, the respondents’ rights, informed consent, the objective of the study, the techniques of the interview and the completion of the questionnaire. The completeness of the data was checked by the data collectors during the data collection and immediately after the data collection by the supervisor and principal investigator.
Data analysis
The generated data were entered into EpiData version 3.14.6 and exported to STATA version 17 for analysis. Categorical variables are presented as frequencies and percentages. Odds ratios with 95% confidence intervals were used to measure the strength of associations. Bivariable logistic regression was performed to determine the crude associations of the independent variables with the dependent variables. Variables with P values < 0.25 in the bivariable logistic regression were exported to the multivariable logistic regression model to control for confounding factors and to identify the independent predictors of PROM. There was no collinearity between variables (ranging from 1 to 2.3), and the Hosmer‒Lemeshow test yielded a P value of 0.387, indicating that the assumption was not violated and that the goodness of fit was not met, and statistical significance was confirmed at a P value < 0.05.