1. Introduction
In the UK, approximately one in five women who access maternity services have a body mass index (BMI) in the obese range (BMI ≥ 30 kg/m
2) [
1], with the odds of obesity being up to five-fold higher among women living in areas of highest versus least deprivation [
2]. The short- and long-term risks of maternal obesity are multiple and severe, including gestational diabetes (GDM), pre-eclampsia, maternal and offspring mortality [
3,
4,
5], and a 264% increase in the odds of childhood obesity in offspring [
6], which may be in part due to nutritional epigenetic changes in utero [
7]. However, pregnancy is an opportunity for intervention to improve maternal diet and physical activity (PA) behaviours and limit gestational weight gain (GWG) which can reduce the risk of GDM, hypertensive disorders of pregnancy, and caesarean section, and improve maternal cardiorespiratory fitness [
8,
9,
10,
11]. Providing weight management support can improve women’s diet and PA behaviours during pregnancy and postnatally, and significantly reduce GWG and postnatal weight retention [
12,
13,
14].
The UK National Institute of Health and Care Excellence guidance for maternal diet includes eating five portions of fruit and vegetables a day, and one portion of oily fish a week. It also states that energy needs do not change in the first 6 months of pregnancy and increase only by 200 calories per day in the last 3 months, and recommends daily supplements including folic acid and vitamin D [
15,
16]. UK recommendations for PA in pregnancy include aiming to achieve at least 150 min of moderate-intensity PA each week, muscle-strengthening activities twice a week and breaking up prolonged periods of sedentary time [
17]. There are currently no national GWG guidelines in the UK other than recommendations that women should not try to reduce obesity-related risks by “dieting” (i.e., losing weight) during pregnancy, and they should be encouraged to lose weight after pregnancy [
18]. The USA Institute of Medicine (IoM) GWG guidelines have been widely adopted internationally and recommend that women with an obese preconception BMI should aim for a total GWG between 5 and 9 kg [
19]. Total GWG includes 0.5–2 kg in the first trimester, and a mean weekly GWG of 0.22 kg (range 0.17–0.27 kg) in the second and third trimesters [
19]. Research into maternal obesity often groups all women with a BMI over 30 kg/m
2 together. However, there is a growing body of evidence showing that obesity-associated risks are greater in higher obesity classes [
20,
21,
22], suggesting that the IoM guidelines should also be stratified by obesity class to reflect differences in risk. A meta-analysis of almost 740,000 women living with obesity has suggested that current recommendations may only be applicable to women with class 1 obesity (BMI 30.0–34.9 kg/m
2), and that a lower total GWG of 1 to <5 kg for women with class 2 obesity (BMI 35.0–39.9 kg/m
2) and 0 kg GWG for women with class 3 obesity (≥40 kg/m
2) could improve foetal growth and caesarean delivery outcomes [
23]. More recently, an observational study using routine data for 337,590 women in Belgium reported that a total GWG of between 0 kg (class 1 obesity) and −5 kg (class 3 obesity) was associated with a reduced risk of a range of outcomes including hypertension, emergency caesarean and high- and low-birth-weight outcomes [
24]. An individual patient data meta-analysis of 196,670 women from 25 cohort studies across Europe and North America reported a reduced risk of one or more adverse pregnancy outcomes with GWG between 2 kg and <6 kg for class 1 obesity, a weight loss or GWG of up to 4 kg for class 2 obesity, and GWG between 0 kg and <6 kg for class 3 obesity [
25]. However, caution is needed in interpreting these results due to a lack of prospective evidence to demonstrate the safety of weight loss in pregnancy.
Despite the significantly increased risks associated with obesity and the strong association with deprivation, there are limited data exploring patterns of diet, PA and GWG among women living with obesity in deprived populations. There is also an absence of data which explores whether there are any differences in patterns between obesity classes. These data would enable us to compare patterns with guideline recommendations, identify target areas for improvement and future interventions for women with the highest level of risk and facing the highest levels of inequality. The aim of this study was to explore the patterns of diet and PA behaviours, and GWG among pregnant women living with obesity in a highly deprived region of England. Additionally, we explored whether there were any differences in these patterns between obesity classes.
4. Discussion
This study aimed to explore maternal patterns of diet and PA behaviours, and GWG in a highly deprived population of women, and whether there were any differences in patterns between obesity classes. Overall, this population had suboptimal dietary intakes relating to low levels of consumption of oily fish, fruit and vegetables, wholemeal bread and unrefined breakfast cereals. EE tended to be from light-intensity PA, and habitual household/care and occupational modes of PA. The majority of women gained weight outside of the IoM guideline recommendations, including both inadequate and excessive GWG. The psychosocial analysis identified that this population of women had a negative body image, but paradoxically a high degree of self-efficacy and internal locus of control relating to their weight and related behaviours. When comparing obesity classes, there was a pattern in the data for women with class 3 obesity to have higher intake of bread, sugar-sweetened beverages, sweet snacks and red meat, EE from sedentary-intensity and inactive modes of PA, and a higher proportion of women with inadequate GWG below the IOM guideline recommendations. Women with class 3 obesity also had a lower overall EE, EE from light- or moderate-intensity PA and occupational mode of PA, lower odds of excessive GWG and a more negative body image. There was also some evidence that both diet and PA patterns change over the course of pregnancy when comparing samples 1 (data collection at approximately 20 weeks gestation) and 2 (data collection at approximately 36 weeks gestation).
This study adds to the scant literature on the habitual diet of UK pregnant women with obesity. The lack of adherence to dietary guidelines is consistent with previous reports and highlights the poor-quality diets consumed by pregnant women with obesity in the UK [
31,
47]. However, we have also reported a potential difference in dietary intake between obesity classes, with women in class 3 reporting more suboptimal dietary patterns compared to those in classes 1 and 2. This requires further research to better understand the relationship between dietary intake and severity of obesity to inform future guidelines for improving the health of pregnant women with obesity, especially considering the higher levels of inequality observed among women with class 3 obesity. Given the association between maternal diet in pregnancy and child weight from birth to adolescence [
48], alongside the increased risk of childhood obesity development when mothers have an obese BMI before pregnancy [
6], efforts to improve diet quality among pregnant women living with obesity has potential intergenerational benefits.
Patterns of self-reported PA in this study were similar to previously reported values using PPAQ [
49]. As might be expected, the EE from occupation and transport related PA was lower in sample 2 (where data were collected at 36 weeks’ gestation) compared with sample 1 (data collected at 20 weeks’ gestation). The number of MET-hr/week spent in moderate intensity PA also decreased between samples 1 and 2. It is difficult to compare self-reported MET-hours per week to the UK Guidelines of 150 min of moderate intensity PA per week. It has been suggested that 16 MET h/week of PA equates to 41 min/day of walking [
50]. However, similar to other studies [
51] our data suggests a reduction in PA over the duration of pregnancy, particularly for women with class 3 obesity. This data reinforces the importance of supporting women to maintain both the amount and intensity of PA throughout pregnancy with the aim of preventing PA decline. Interestingly, only a small amount of total EE was expended carrying out sports activities, suggesting that most of the reported moderate intensity activity was not from formal exercise or sporting activities. There is evidence that some women are uncertain of appropriate exercise for pregnancy, with particular concerns relating to safety [
52]. Women should be reassured that they can meet PA guidelines by carrying out habitual activities such as walking and actively playing with children, as well as appropriate sports- and exercise-based activities. Future research involving PA interventions in this population should focus on maintaining amount and intensity of habitual PA throughout pregnancy to help achieve national guideline recommendations, as well as exploring potential barriers to PA in pregnancy among women in different classes of obesity.
The data on GWG demonstrated a lack of adherence to the IOM-recommended GWG ranges for all women, with obesity class apparently influencing whether the GWG was inadequate or excessive. Other studies have demonstrated that GWG tends to decline, and the risk of weight loss increases, with increasing obesity class [
53]. Our study provides novel data in the extent to which women with class 3 obesity appear to have a different overall patterns of GWG to women in classes 1 and 2. While there is a lack of conclusive mechanistic evidence as to why this pattern may have been observed, it may be related, in part, to resting energy expenditure (REE) which accounts for between 50 and 75% of total EE. In non-pregnant populations, REE is approximately 360 kcal/day higher among people living with obesity than for those without obesity, with differences observed according to obesity class: 240 kcal/day higher for class 1 obesity, and 540 kcal/day higher for class 3 obesity [
54]. A study in pregnant women living with obesity also identified that increasing obesity class was significantly associated with increasing REE (obesity class 1: 1686 ± 39 kcal/day; class 2: 1775 ± 50 kcal/day; class 3: 2102 ± 58 kcal/day) [
55]. A key contributor to REE is fat-free mass (FFM), although fat mass (FM) also contributes, and both are higher in people living with obesity than for those without obesity [
54]. In pregnancy, increasing obesity class is significantly associated with both FFM (class 1: 50.4 ± 1.1 kg; class 2: 54.0 ± 1.5 kg; class 3: 60.7 ± 2.1 kg) and FM (class 1: 36.8 ± 1.0 kg; class 2: 44.8 ± 1.1 kg; class 3: 62.7 ± 2.9 kg) [
55]. Further, REE increases over the course of pregnancy [
56,
57], and change in REE has been identified as being an important factor in determining GWG [
58]. In a study with pregnant women (non-obese), change in REE correlated positively with changes in FFM and negatively with FM, and women with smaller increases in REE had higher GWG [
58]. However, there is a lack of data exploring the extent to which change in REE differs between obesity classes. Based on evidence to date, it could be hypothesised that the lower GWG observed in the class 3 obesity group in GLOWING could be due to this population having a higher REE and increase in REE over the course of pregnancy, and a higher accrual of FFM than FM, compared with women with class 1 or 2 obesity. As such, the energy intake required for women with class 3 obesity to meet their REE demands, plus pregnancy-specific energy demands, and to gain excessive weight during pregnancy may be too high for some women, resulting in inadequate GWG. It is also important to note that our study applied the recommended GWG ranges for women with pre-pregnancy obesity (BMI ≥ 30.0 kg/m
2) in order to explore adherence to guideline recommendations. This included a combination of first trimester GWG, and weekly GWG for the second and third trimesters. However, the IoM GWG guidelines do not differentiate between obesity classes in their recommendations for any of these trimester-specific GWG ranges. As discussed in the introduction to this paper, recent studies to date suggest that obesity class is an important factor in determining optimal GWG [
23,
24,
25]. Collectively, these studies suggest that women in obesity classes 1–3 have different GWG requirements for prevention of adverse pregnancy outcomes, although there is disagreement on what these optimal ranges for each obesity class should be. The optimal GWG is reported by authors to be 5 kg to 9 kg [
23], 0 kg [
24], and 2 kg to <6 kg [
25] for class 1 obesity; 1 kg to <5 kg [
23], −4 kg [
24], and “weight loss” or a GWG < 4 kg [
25] for class 2 obesity; 0 kg [
23], −5 kg [
24] and 0 to <6 kg [
25] for class 3 obesity. This provides further evidence for the need to treat obesity classes as separate populations in future research, guidelines and practice, as women living with obesity are not a homogeneous group.
In this study, we employed a novel method of determining adequacy of gestation-specific GWG informed by a published comparison of approaches [
38]. Our approach used the latest weight measurement recorded in the second and third trimesters in order to optimise sample size (
n = 90) due to the lack of consistency in gestational age at time of weight measurement, and the lack of preconception and delivery weight measurements which are required to apply the IoM total GWG guidelines [
52]. As a sensitivity analysis, we compared these results to the GWG categorisations of a subgroup of women with weight recorded at 36 weeks’ (
n = 51) to estimate the extent to which inconsistency and timing of measurement impacted on the results. We found that the results were highly comparable in both frequency and regression analyses which suggests that our approach of applying gestation-specific GWG ranges to weights recorded in the second or third trimester could be used to maximise sample size in future studies. We also performed multiple imputation to estimate weight at 36 weeks for all women (
n = 163) and compared these with our complete case analyses. This showed generally similar trends in excessive GWG across obesity classes, but notably the effect size for excessive GWG reduced in those with class 3 vs. class 1 obesity when compared to complete cases analysis (OR = 0.34 vs. OR = 0.12). While this could be closer to the ‘true’ effect size, caution should be taken interpreting the results from the multiple imputation given the small sample size and the high proportion of missing weight data at 36 weeks (69%). While we were able to perform 20 iterations and include multiple important variables in the imputation, the limited complete case sample size meant we could not reasonably include all variables that might be important to predict 36 week weight (e.g., diet and PA data).
The psychosocial scales demonstrated negative body image, while also a high degree of internal locus of control (i.e., belief in their ability to control their own weight) and self-efficacy (i.e., confidence that they can control their weight and related behaviours). Most women also reported that they had successfully tried to lose weight in the past, which may be influencing the high degree of self-efficacy and internal locus of control. However, this pattern in the data may indicate internalised weight bias in this population, whereby women feel that the negative stereotypes about obesity causation (e.g., lack of will power, greed, laziness) apply to themselves [
59]. In this context, the high internal locus of control may reflect that women feel they should have complete control of their weight and therefore having a high weight represents their failure in control (reflected in the negative body image scale). Others have identified that perceived weight stigma and internalised weight bias should be considered in the design and evaluation of interventions to improve mental health among people living with obesity [
60]; however, there is a lack of research among pregnant populations.
Strengths and Limitations
The data collection for the GLOWING pilot trial was extensive. The volume of data collected provides a rich dataset for analysis of patterns of diet, PA and GWG among a high risk and socio-economically disadvantaged population of women. However, as this was a secondary analysis of pilot trial data, the research was not designed or powered to detect statistically significant differences in the analysis we carried out. Due to this study being an exploratory analysis of a rich dataset on diet, PA and GWG, with data collection at different gestations and among women with different obesity classes, we have carried out multiple testing. The small sample size means that any statistically non-significant results does not necessarily mean there is no association between obesity classes and the outcomes explored, rather that this study is likely to be underpowered to detect a difference. As the aim of this study was to explore the patterns of diet and PA behaviours, and hereby identify areas for future research, we have not adjusted our findings for multiple testing [
61], this is a similar approach taken by other analyses of trial data [
62]. Therefore, the results of this study are descriptive, and we have primarily focused our discussion on the clinical significance relating to patterns in the data rather than statistical significance, with the purpose of identifying potential areas for future research.
The data were collected from women representing a highly deprived population living in a high-income country. The socio-economic data were collected at both the area-level (i.e., IMD), and the individual-level (i.e., maternal education and employment). There was consistency in all of these indicators with a higher proportion of women in this sample living in areas of highest deprivation, having low levels of education and high rates of unemployment compared with national averages. There was also evidence that women living with class 3 obesity in this population had higher levels of some inequality measures than women living in the same geographic area with class 1 or 2 obesity, including measures relating to deprivation quintile, employment and education. However, this population was predominantly White, and women who did not speak English were excluded from this study, which will impact on how comparable our dataset is to other regions in the UK. There is a plethora of evidence which identifies that the social gradients in high income countries can result in significantly increased risks for women and their babies in the low socio-economic groups [
63]. Further, all women in this study were at additional high risk due to their BMI being in the obese range, and have the greatest potential for benefit from interventions. We have very little existing evidence on patterns of diet, PA and GWG in populations of highly deprived women living with obesity, and this study provides a basis to inform future study direction.
This exploratory study has identified multiple areas for future research, policy and practice. Women living with obesity and in areas with high levels of deprivation have a double burden of disease, as both obesity and deprivation increase risks of adverse pregnancy outcomes. This population of women require additional support to achieve guideline recommendations for diet, PA and GWG. Future research should explore the differences between obesity classes in both diet and PA behaviours and GWG to inform the development of guidelines; currently, diet and PA guidelines in the UK are not specific to women living with obesity and there are no UK GWG guidelines. The patterns in the psychosocial constructs and the potential role of internalised weight bias relating to maternal diet, PA and GWG should also be further explored. Further research into maternal diet, PA and GWG among obesity classes, with a particular focus on the needs of deprived populations, would inform the development of appropriate policies and provision of care to achieve the best health outcomes for women and their children.