Study population
This national prospective longitudinal study recruited pregnant women living in the 6 different governorates of Lebanon: Mount Lebanon, Beirut (capital of Lebanon), Bekaa, South (and Nabatieh), North, and Akkar. A simple random sampling among all obstetric clinics in Lebanon was done using the statistical software SPSS (Statistical Package for Social Sciences), version 22.0. Of the 732 obstetric clinics, a total of 20 private and hospital based private clinics were selected using simple random sampling: Mount Lebanon (n = 3), Beirut (n = 5), Bekaa (n = 4), South (n = 3), North (n = 4), and Akkar (n = 1).
A systematic approach was used to recruit approximately 30 women per clinic. Clinic staff generated a list of all eligible patients attending prenatal visits during the recruitment period, and participants were selected using a simple random sampling approach. Women were approached in person, provided with detailed study information, and asked to provide written informed consent. Recruitment was conducted between 2019 and 2021 and coincided with the COVID-19 outbreak and Lebanese economic crisis. Consenting pregnant women were followed from the 1st trimester (<15 weeks of gestation) until delivery. The initial recruitment interview was conducted in person with follow-up interviews by phone in the second (24–28 weeks), and third trimesters (34–37 weeks).
We used Epi Info Software to calculate the sample size needed using the formula by Fleiss with correction [15]. The confidence interval was set at 95%, power at 80% and ratio (unexposed/exposed to the MeD=1). A sample of 618 participants was targeted after adjusting for a 20% loss to follow up. Of the 660 participants, 42 dropped out due to miscarriages and/or other reasons.
Inclusion criteria were pregnant Lebanese women > 18 years with a singleton pregnancy. Exclusion criteria included women with multiple gestations, or pre-pregnancy diabetes, or those carrying a fetus with structural malformation, chromosomal anomalies or TORCH (toxoplasmosis, rubella, cytomegalovirus, herpes and other agents) infections.
Study design
A survey was administered which included a section on socio-demographic data. Validated questionnaires were used to assess stress [16], sleep [17], depression [18], adherence to the LMeD [13] and physical activity [19] at T1, T2 and T3. Clinical data was obtained through medical charts, including self-reported pre-pregnancy weight, vitamin/mineral intake, presence of anemia and medical history (maternal previous GDM and family history of diabetes). At follow-up visits at each trimester, additional clinical indicators were collected from medical charts including IGT per trimester, diagnosis of GDM, and general maternal health status (anemia, vitamin/mineral supplementation, COVID). The main exposure variable was LMeD adherence which was assessed in each trimester. The main outcomes were: (1) diagnosis of GDM between 24 and 28 weeks and (2) IGT in T1 and T3. IRB approval was obtained from Bellevue Medical Center’s institutional review board in accordance with the 1964 Helinski Declaration [20]. Written consent was obtained from each physician at the 20 private clinics in order to recruit their patients and access medical records.
QuestionnairesDietary assessment and adherence to the Mediterranean diet
Food intake was collected using a validated 61-item food frequency questionnaire (FFQ) for the Lebanese diet as it incorporates traditional Lebanese dishes [13]. This FFQ consists of 9 food categories including (1) breads and cereals, (2) dairy products, (3) fruits and juices, (4) vegetables, (5) meat and alternates, (6) fats and oils, (7) sweets and desserts, (8) beverages, and (9) miscellaneous.
This LMeD pattern consisted of 9 food groups that had high loading on this pattern which included: fruits, vegetables, legumes, olive oil, burghol (crushed whole wheat), milk and dairy products, starchy vegetables (potato, corn and beans), dried fruits and eggs [13]. Adherence to the LMeD was measured based on the consumption of each of the nine food groups with 1, 2 and 3 points assigned to intakes in the 1st, 2nd and 3rd tertile indicating low, medium and high intake, respectively. Scores ranged between 9 and 27, with 9 reflecting the least adherence and 27 reflecting the highest adherence to the LMeD [13]. Alcohol intake was excluded as consumption is virtually absent in this study population. This tool showed a good correlation (r = 0.56, p < 0.01) with the Italian MeD adherence tool [21].
Edinburgh perinatal/postnatal depression scale (EPDS)
Edinburgh Perinatal/Postnatal Depression Scale (EPDS) is a commonly used 10-item scale in clinical practice to identify women at risk of perinatal depression. A global sum of 0-30 is obtained. A score of 10 or greater indicates possible depression [18]. It has been validated in a Lebanese pregnant population with Cronbach’s alpha=0.84, correlation (r = 0.57, p < 0.01) with the Present State Examination [22].
Perceived stress scale (PSS10)
Perceived stress was evaluated using the Perceived Stress Scale (PSS10), which was later shortened to 10 items [16]. It has been validated in a Lebanese pregnant population with test-retest reliability and Cronbach’s alpha=0.74; Correlation r = 0.48, p < 0.001 and r = 0.58, p < 0.001 with a general health questionnaire and EPDS, respectively [14]. A global sum score of 0–40 is obtained with higher scores indicating higher perceived stress. The categories are: Low stress (score 0–13), moderate stress (score 14–26), and high stress (score 27–40).
Pittsburgh Sleep Quality Index (PSQI)
PSQI is a self-reported tool that assesses sleep quality over a 1-month period. A global score ranges between 0 and 21. A score of equal or greater to 5 indicates poor sleep, while a score of less than 5 indicates good sleep. It has been validated in an Arabic population with Cronbach’s alpha=0.74, correlation r = 0.36, p < 0.001 and r = 0.19, p = 0.03, with the Insomnia Severity Index and Medical Outcome Study Short Form, respectively [17].
Canadian Physical Activity Readiness Examination (PARmed-X)
PARmed-X was used to assess physical activity in the past month [19]. Participants were categorized according to the frequency, intensity and duration of physical activity. Accordingly, categories were “sedentary” (frequency <1-2 times/ week and <20 min duration; PA index=0), and “active” (1-2 times/ week for 20 min or more than twice/week for < 20 min; PA index=1 or >2 times/week for more than 20 min; PA index=2).
Clinical measurementsPre-pregnancy Body Mass Index (BMI) and Gestational Weight Gain (GWG)
Pre-pregnancy BMI was collected at baseline from obstetrical charts and was used to classify women as underweight (<18.5 kg/m2), normal (18.5–24.9 kg/m2), overweight (25-29.9 kg/m2) or obese (≥30 kg/m2). Total GWG in Kg was calculated by subtracting pre-pregnancy weight from weight at 39 weeks gestation, and GWG was classified as low, adequate or excessive according to the pre-pregnancy BMI using the Institute of Medicine (IOM) recommendations (underweight 12.5-18 kg, normal weight 11.5-16 kg, overweight 7-11.5 kg, and obese 5-9 kg). GWG per trimester was also calculated and classified as adequate, excessive or low according to the following recommendations: For 1st trimester: above 2 kgs or below 1 kg was considered excessive or low, respectively for all categories of BMI. For 2nd and 3rd trimester, the recommended weight gain per week for each BMI category is the following: 0.45–0.6 kg/wk for underweight women, 0.35–0.45 kg/wk for normal weight women, 0.2–0.3 kg/wk for both overweight, and obese women. Weight gain below or above these values was classified as low or excessive, respectively [23].
FBG, IGT and GDM
FBG values were collected from obstetric charts at T1 and T3, and were categorized into normal FBG: < 100 mg/dl and IGT ≥ 100 mg/dl, while the presence of GDM diagnosis (yes/no) was confirmed during the visit in the second trimester [24]. Obstetricians in Lebanon followed the two-step 50 g oral glucose tolerance test between 24-28 weeks in line with the American College of Obstetricians and Gynecologists [25]. If blood glucose was ≥140 mg/dL one hour after the intake of 50 g glucose, a 100 g three-hour oral glucose tolerance test was performed. In the latter, blood glucose was measured at t = 0 (fasting), t = 1 h, t = 2 h and t = 3 h with cut-off values (Carpenter and Coustan criteria) of 95, 180, 155, and 140 mg/dL, respectively [23]. GDM diagnosis was made when two or more blood glucose levels met or exceeded the cut-off values [24].
Mean Arterial Pressure (MAP)
MAP is a novel blood pressure index that was more sensitive in predicting abnormal blood pressure in this population. It is computed using systolic and diastolic blood pressures (SBP and DBP, respectively) and can be elevated despite mothers having SBP AND DBP within normal ranges. MAP is calculated using the formula. MAP = DBP + 0.33 [SBP-DBP] (25). Trimester-specific cutoffs for elevated MAP (eMAP) in pregnancy are defined as >87 mmHg (10- < 18 weeks), >84 mmHg (18-34 weeks), and >86 mmHg (after 34 weeks), while low MAP is defined as <70 mmHg [26].
Statistical analyses
Univariate analyses (measures of central tendency, percentage and frequencies) were used to summarize the population characteristics of pregnant women as well as the frequency of consumption of each food group of the LMeD. Diet, GWG, smoking, physical activity, and psychosocial factors including stress, sleep and depression were compared across the trimesters using the McNemar test. Frequencies and percentages were used to determine the categorical dietary variable: adherence to the LMeD, while means were used to determine mean adherence of the population as well as mean intake of each of the 9 food groups of the LMeD. Scatter plots were used to describe the FBG measurements in trimester 1 and 3. Maternal characteristics were also compared among women with normal FBG (<100 mg/dl) vs. IGT (≥100 mg/dL) in trimesters 1 and 3.
Hierarchical logistic regression models were used to test the associations between predictors and elevated FBG in trimesters 1 and 3, while in trimester 2, models were used to test the associations between GDM and the same predictor variables. In model 1, GWG, pre-pregnancy BMI, family history of diabetes, stress, sleep, depression and MAP were entered. In model 2, adherence to the LMeD was added, and in model 3, each of the 9 food groups in individual models along with the significant variables from model 2 were entered. All the analyses were conducted using SAS V9.4.
Dining and Cooking