1 INTRODUCTION

Early-life exposures starting from preconception have been increasingly recognized to play an important role in short- and long-term health outcomes in both mothers and offspring, consistent with the Developmental Origins of Health and Disease hypothesis.1 Prepregnancy body mass index (BMI) and gestational weight gain (GWG) represent important physical markers of a woman's overall living conditions and lifestyle behaviors throughout pregnancy. A large body of evidence has demonstrated that higher prepregnancy BMI and GWG above the Institute of Medicine (IOM) guidelines2 are associated with higher risks for pregnancy complications including gestational hypertension, gestational diabetes mellitus, caesarean delivery, and large for gestational age at birth.3-5 These associations remain similar across continents and ethnicities even when population-specific BMI categories were considered.6 Moreover, preconception overweight and obesity is associated with poorer neurocognitive development7 and childhood overweight and obesity in offspring.8 These findings suggest that the need to develop interventions targeting prepregnancy BMI and GWG to reduce the burden of possible future health issues in both the mother and child.

Diverse factors have been associated with overweight and obesity, and excessive GWG, which include physiological, psychological, environmental, social, and behavioral exposures,9, 10 and notably, these factors usually vary by sociodemographic status.11 It is possible that sociodemographic determinants may influence individual choices or preferences of exposures and behaviors, subsequently affecting prepregnancy BMI and GWG. A systematic review found that low maternal education, rather than other sociodemographic factors such as income and employment, tended to be associated with excessive GWG.12 However, previous findings were mainly limited to developed Western settings,12 which may not be generalizable to other regions with different sociodemographic backgrounds and cultural behaviors. Moreover, previous studies rarely included different aspects of sociodemographic status relevant to other demographics, such as household occupancy.

The United Arab Emirates (UAE) is a high-income Arab country with a relatively homogeneous native Emirati population.13 Previous research has shown that the Middle East region has a high prevalence of overweight and obesity, especially amongst women.14-16 Studies in the UAE have shown that approximately two-thirds of Emirati women were classified as overweight or obesity; however, these cross-sectional designs were unable to elucidate the risk factors for weight gain.17, 18 Due to the cultural practices and religious beliefs of women in the UAE, the prevalence of tobacco smoking19 and alcohol consumption is extremely low.17 However, a birth cohort of Emirati and Arab women in the UAE reported that more than half of women with overweight or obesity, and almost three-quarters of pregnant women had inadequate and excessive GWG.20 This study did not find any association between maternal education, employment status, or monthly family income and prepregnancy BMI or GWG among the UAE pregnant women; however, the sample size was small (N = 256) and the analysis may have been underpowered for this specific analysis.20 A recent systematic review on maternal and child cohorts in the Gulf Cooperation Council countries including the UAE highlighted a lack of research on GWG in the region.21 As can be seen by previous literature mentioned above, overweight and obesity is an important public health issue in the UAE, and its effects on GWG in pregnancy needs to be thoroughly understood using the robust longitudinal cohort design with a large representative sample. Hence, this study aims to investigate the associations between sociodemographic and lifestyle factors and prepregnancy BMI and GWG in the largest mother and child prospective cohort study in the UAE. It was hypothesized that women with overweight or obesity before pregnancy would be more likely to have poorer lifestyle factors and unhealthy GWG during their current pregnancy.

2 METHODS 2.1 Study participants

The Mutaba'ah (meaning “follow-up” in Arabic) study is an ongoing prospective cohort study in Al Ain city, UAE, that plans to follow the mothers and their offspring until the child turns 16 years of age.22 Since 2017, pregnant women (at any week of gestation) have been recruited from the three major hospitals in Al Ain (7690 pregnant women were recruited as of November 2020). The recruitment criteria included: women from the Emirati population resident in Al Ain, at least 18 years old, inclusion of their newborn(s) and being able to provide informed consent. Participants’ information was ascertained from medical records and using tablet-assisted self-administered questionnaires in Arabic. Further details of the study design have been described elsewhere.22 Ethical approval for this cohort study was obtained from the UAE University Human Research Ethics Committee (previously known as Al Ain Medical District Human Research Ethics Committee) (ERH-2017-5512), Al Ain Hospital Research Ethics Committee (AAHEC-03-17-058), and Tawam Human Research Ethics Committee (T-HREC-494).

2.2 Sociodemographic and lifestyle factors

A wide range of indicators including maternal education, occupation (i.e. employment status) and income (i.e., type of housing, housing ownership, and number of residents in household) reported at the recruitment visit were used to represent different aspects of sociodemographic status.23 While education might indicate both knowledge-related assets and economic resources and status, occupation might reflect status, prestige, or community ranking.24, 25 Income related variables might represent wealth or material aspects of sociodemographic status and assets.24, 25

Maternal highest education was classified into: 1) none/primary/secondary education, 2) postsecondary education including vocational and diploma degree and 3) tertiary education including bachelor, master, and doctoral degrees. Responses on employment status were recorded as: 1) student/unemployed/retired, 2) housewife, or 3) employed/self-employed.

Pregnant women's housing type (standalone homes [these are government initiative housing] apartment, part of a villa, and villa) and ownership (rented, owned) was used as an economic indicator. Pregnant women were asked to indicate the number of residents living in their household, which were categorized into tertiles (≤6, 7–12, and ≥13 people). Parity was also included as part of sociodemographic measures, classified into nulliparous, multiparous (1–2, 3–4 children) and grand multiparous (≥5 children).

Lifestyle factors that are likely to precede pregnancy were also included in the analyses. Pregnant women provided binary responses (yes or no) on pregnancy planning status and infertility treatment status at recruitment visit. Prepregnancy maternal and husband's active smoking status, ranging from never to regularly, were also recorded.

2.3 Prepregnancy BMI and GWG

Information on maternal age and on repeated measures of maternal weight before and during pregnancy and maternal height were extracted from medical records. Prepregnancy maternal weight was defined as the most contemporaneous body mass between eight weeks before pregnancy and first month of pregnancy.

Since gestational weight was serially measured across a wide range of gestational ages, the estimation of total GWG is subject to varying gestational age intervals between the first and last measures. Therefore, random coefficient modeling of weight measures during pregnancy between first and third trimester was computed. This model generates the individual Best Linear Unbiased Predictor values of the random intercept (i.e., the difference between the person-specific intercept and the overall intercept) and random slope (i.e., the difference between person-specific slope and the overall slope).26 The individual linear trajectories of GWG per week from gestational age at 8 weeks to delivery was then estimated by adding the overall slope to the individual random slopes. Similarly, prepregnancy weight was estimated by summing the overall intercept and the individual random intercepts from the same model. Given the high correlation between the estimated and the measured prepregnancy BMI (Pearson's correlation coefficient = 0.97), the estimated prepregnancy weight was used for women with missing (n = 1658) data on the prepregnancy weight.

The GWG per week was categorized into i) inadequate, ii) adequate, or iii) excessive for a given prepregnancy BMI status, according to the IOM's guidelines.2 The adequate range of GWG per week was considered as: 0.44–0.58 kg for women with underweight, 0.35–0.50 kg for women with normal weight, 0.23–0.33 kg for women with overweight, and 0.17–0.27 kg for women with obesity.2 GWG below or above these weight gain ranges for a given prepregnancy BMI status were considered as an inadequate or excessive gain, respectively.

2.4 Statistical analyses

This study included pregnant women who had delivered and had data on both prepregnancy weight status and rate of GWG. The differences in socioeconomic determinants across prepregnancy BMI and GWG adequacy were compared using chi-squared tests for categorical variables and one-way analysis of variance tests for continuous variables.

To explore associations between sociodemographic determinants and prepregnancy BMI (as a continuous outcome), multivariable linear regression models with adjustment for maternal age at delivery were computed. The regression coefficients (β) with their 95% confidence interval (CI) were reported to demonstrate the unit change of BMI associated with changes in each of the sociodemographic and lifestyle factors. Similarly, the associations between sociodemographic determinants and GWG adequacy were tested using multivariable multinomial logistic regression, with adequate GWG as the reference category. For logistic regression models, odds ratios (ORs) with 95% CI were reported. These models were conducted i) separately in women with prepregnancy normal, overweight, and obesity weight status, adjusted for maternal age at delivery and ii) in all women, adjusted for maternal age at delivery, and prepregnancy BMI. Further adjustment by including all sociodemographic determinants in the same models were performed. To allow comparisons of findings across sociodemographic determinants, multiple imputation by chained equations with 50 datasets were performed to impute missing sociodemographic determinants (n = 221–428, 6.3%–12.1%).

All statistical analyses were performed using Stata 15.1 (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC). Statistical significance was defined by a p ≤ 0.05 and 95% confidence intervals.

3 RESULTS 3.1 Participants' characteristics

As of November 2020, 4399 pregnant women had given birth and 3536 (80.4%) pregnant women were included in the present analysis. Those excluded were due to missing data on both prepregnancy BMI and GWG (n = 854) or on only GWG (n = 9). Compared to the excluded population, included women were older, multiparous, lower educated, and housewives. They were also more likely to live in a rented apartment, have planned their pregnancy, have had infertility treatment, and have a husband who did not smoke before pregnancy (Supplementary Table S1).

Among the included pregnant women, the mean age at recruitment and the mean prepregnancy BMI of pregnant women were 31.4 ± 6.1 years old and 26.8 ± 5.9 kg/m2, respectively. Based on the prepregnancy BMI, one-third (33.5%) of pregnant women were classified as normal weight, followed by one-third (33.2%) classified as overweight, 26.9% as obesity, and 6.4% as underweight. More than one-quarter (27.6%) of pregnant women were considered to have an adequate rate of GWG, while approximately one-third (34.2%) each had an inadequate or excessive (38.2%) rate of GWG. Women with overweight and obesity were more likely to have excessive GWG rate (49.9% and 48.8% vs. 18.6% and 21.8%) than women with underweight and normal weight (Figure 1).

Gestational weight gain stratified by prepregnancy BMI status among pregnant women in Al Ain, UAE: The Mutaba'ah study

Table 1 shows the comparisons of pregnant women's characteristics across prepregnancy BMI and GWG adequacy. Compared to women with underweight (mean age ± SD: 26.8 ± 4.6 years) and normal weight (29.7 ± 5.6 years), women with overweight (32.1 ± 5.9 years), and obesity (33.6 ± 5.8 years) tended to be older. They also had a greater prevalence of grand multiparity (19.4% [overweight] and 24.6% [obesity] vs. 3.9% [underweight] and 11.3% [normal weight]), and lower proportions of them were educated (tertiary: 39.2% and 34.0% vs. 41.7% and 45.4%), employed or self-employed (34.7% and 37.1% vs. 15.0% and 29.2%), and more frequently had previous infertility treatment (12.3% and 14.8% vs. 5.2% and 8.4%) (Table 1). Furthermore, pregnant women with excessive GWG rate were less likely to be grand multiparous (14.5% vs. 15.7% [adequate] and 20.7% [inadequate]) but more likely to be younger (mean age: 30.8 ± 6.0 vs. 31.1 ± 6.0 and 32.1 ± 6.1 years) and have planned their pregnancy [59.4% vs. 56.2% and 54.2%], than their counterparts with adequate or inadequate GWG rate, respectively.

TABLE 1. Comparisons of characteristics across pregnant women's prepregnancy BMI status and gestational weight gain adequacy among pregnant women in Al Ain, UAE: The Mutaba'ah study Characteristics Prepregnancy BMI Status Gestational weight gain rate Underweight (n = 226) Normal (n = 1186) Overweight (n = 1174) Obesity (n = 950) p value Inadequate (n = 1210) Adequate (n = 975) Excessive (n = 1351) p value Mean ± SD Mean ± SD Mean ± SD Mean ± SD Mean ± SD Mean ± SD Mean ± SD Age at delivery, years 26.8 ± 4.6 29.7 ± 5.6 32.1 ± 5.9 33.6 ± 5.8 <0.001 32.1 ± 6.1 31.1 ± 6.0 30.8 ± 6.0 <0.001 Prepregnancy body mass index, kg/m2 16.9 ± 1.3 22.1 ± 1.8 27.4 ± 1.4 34.2 ± 4.0 <0.001 26.1 ± 6.5 25.8 ± 5.6 28.1 ± 5.3 <0.001 Gestational weight gain per week, kg 0.44 ± 0.16 0.39 ± 0.16 0.34 ± 0.18 0.27 ± 0.19 <0.001 0.18 ± 0.13 0.34 ± 0.10 0.49 ± 0.14 <0.001 n (%) n (%) n (%) n (%) n (%) n (%) n (%) Parity    <0.001 <0.001 0 89 (43.0) 307 (29.4) 164 (15.9) 108 (13.1) 188 (17.5) 193 (22.5) 287 (24.5) 1–2 87 (42.0) 381 (36.4) 351 (34.0) 253 (30.8) 361 (33.6) 296 (34.5) 415 (35.4) 3–4 23 (11.1) 240 (22.9) 318 (30.8) 259 (31.5) 303 (28.2) 235 (27.4) 302 (25.7) ≥5 8 (3.9) 118 (11.3) 200 (19.4) 202 (24.6) 223 (20.7) 135 (15.7) 170 (14.5) Maternal education    <0.001 0.081 None/primary/secondary school 107 (50.7) 499 (45.2) 560 (51.7) 485 (55.3) 595 (52.8) 428 (47.0) 628 (50.7) Postsecondary 16 (7.6) 104 (9.4) 98 (9.1) 94 (10.7) 109 (9.6) 85 (9.3) 119 (9.6) Tertiary 88 (41.7) 502 (45.4) 425 (39.2) 298 (34.0) 423 (37.6) 398 (43.7) 492 (39.7) Employment    <0.001 0.309 Student/unemployed/retired 68 (31.9) 209 (18.9) 137 (12.7) 81 (9.2) 155 (13.8) 151 (16.6) 189 (15.2) Housewife 113 (53.1) 573 (51.9) 568 (52.6) 471 (53.6) 616 (54.8) 459 (50.5) 650 (52.4) Employed/self-employed 32 (15.0) 322 (29.2) 375 (34.7) 326 (37.1) 354 (31.5) 299 (32.9) 402 (32.4) House type    0.156 0.487 Rented flat/apartment/standalone home 15 (7.2) 75 (7.0) 60 (5.7) 55 (6.4) 74 (6.7) 56 (6.4) 75 (6.1) Rented part of a villa/villa 18 (8.6) 156 (14.6) 133 (12.6) 124 (14.4) 158 (14.4) 127 (14.4) 146 (12.0) Owned flat/apartment/standalone home 25 (12.0) 148 (13.8) 173 (16.3) 133 (15.5) 167 (15.2) 135 (15.3) 177 (14.5) Owned part of a villa/villa 151 (72.3) 693 (64.7) 694 (65.5) 548 (63.7) 701 (63.7) 562 (63.9) 823 (67.4) Number of residents in household    0.127 0.054 ≤6 68 (33.0) 307 (28.9) 309 (29.6) 259 (30.5) 339 (31.4) 228 (25.9) 376 (31.3) 7–12 66 (32.0) 376 (35.4) 415 (39.8) 305 (36.0) 385 (35.6) 337 (38.3) 440 (36.6) ≥13 72 (35.0) 380 (35.8) 320 (30.7) 284 (33.5) 357 (33.0) 314 (35.7) 385 (32.1) Planned pregnancy    0.886 0.035 No 97 (45.5) 472 (42.7) 473 (43.2) 387 (43.6) 522 (45.8) 395 (43.8) 512 (40.6) Yes 116 (54.5) 634 (57.3) 623 (56.8) 501 (56.4) 618 (54.2) 507 (56.2) 749 (59.4) Infertility treatment    <0.001 0.160 No 199 (94.8) 1005 (91.6) 945 (87.7) 746 (85.2) 1012 (90.1) 792 (88.7) 1091 (87.6) Yes 11 (5.2) 92 (8.4) 133 (12.3) 130 (14.8) 111 (9.9) 101 (11.3) 154 (12.4) Prepregnancy maternal active smoking 0.052 0.209 Never 216 (100.0) 1095 (98.8) 1085 (98.1) 865 (97.7) 1124 (98.4) 903 (98.9) 1234 (97.9) Occasionally/regularly 0 (0.0) 13 (1.2) 21 (1.9) 20 (2.3) 18 (1.6) 10 (1.1) 26 (2.1) Prepregnancy husband's active smoking 0.616 0.947 Never 133 (61.6) 664 (59.8) 647 (58.7) 520 (58.9) 672 (58.8) 548 (60.3) 744 (59.1) Occasionally 38 (17.6) 182 (16.4) 197 (17.9) 136 (15.4) 190 (16.6) 147 (16.2) 216 (17.1) Regularly 45 (20.8) 264 (23.8) 258 (23.4) 227 (25.7) 280 (24.5) 214 (23.5) 300 (23.8) 3.2 Sociodemographic factors and prepregnancy BMI

Supplemental Table S2 and Figure 2 show the adjusted associations of sociodemographic and lifestyle factors with prepregnancy BMI. Higher parity (β for 1–2 (vs. none): 0.92 kg/m2, 95% CI: 0.36–1.48; 3–4: 1.48 kg/m2, 95% CI: 0.84–2.012; ≥5: 1.71 kg/m2, 95% CI: 0.88–2.54), having previous infertility treatment (β: 0.69 kg/m2, 95% CI: 0.08–1.30), prepregnancy maternal active smoking (β: 1.95 kg/m2, 95% CI: 0.42–3.49) and prepregnancy husband's active smoking (β: 0.53 kg/m2, 95% CI: 0.07–1.00) were linearly associated with higher prepregnancy BMI (p < 0.05). Additionally, higher levels of maternal education were linearly associated with lower prepregnancy BMI (β for tertiary: −1.42 kg/m2, 95% CI: −1.82 to −1.02). Owning compared to renting a flat was associated with higher prepregnancy BMI (β: 0.93 kg/m2, 95% CI: 0.03–1.82). Similar findings were seen in the mutual adjustment model except for prepregnancy husband's active smoking and house type (Supplementary Table S2).

Factors associated with prepregnancy body mass index in pregnant women in Al Ain, UAE: The Mutaba'ah study

3.3 Sociodemographic factors and GWG

Table 2 shows the adjusted associations of sociodemographic and lifestyle factors with GWG rate classification by prepregnancy BMI status. Among women with normal