The selection process is illustrated in the flow diagram shown in Fig. 1. The database search returned 7,700 records. After removing the duplicates, 5,081 articles were screened for eligibility. A total of 4,903 studies were excluded based on title and abstract, and 108 were excluded based on full-text screening, resulting in 57 studies being included. In addition, we tracked the article references and added 13 articles. Finally, 70 studies were included in the qualitative synthesis and 33 in the quantitative synthesis.
Basic characteristics of the included studies
The basic information for each included study is shown in Additional File 2. Seventy studies were published from January 2009 to November 2020, including 48 cohort studies (68.6%), 7 case–control studies (10.0%), and 15 cross-sectional studies (21.4%); 10% of them were in Chinese (N = 7) and the rest were in English. The included studies were conducted in Americas (n = 33), Asia (n = 22), Europe (n = 11), Oceania (n = 3), and Africa (n = 1).
In the 48 cohort studies, 3,053,581 women were investigated worldwide, 1,473,843 of them had EGWG, and the prevalence of EGWG was approximately 45.5% (95% CI:42.8%, 48.3%). Among the different regions, the prevalence of EGWG was highest in Americas, 49.8% (95% CI, 46.9% − 52.8%) and Africa (55.20%), followed by Oceania at 49.0% (95% CI, 17.9% − 80.2%) and Europe (42.3% 95% CI:35.9% − 48.6%), and the lowest prevalence in Asia at 38.2% (95% CI, 27.8% − 48.6%).
Quality of the literature
For full quality appraisal, see Additional File 3. The quality of all included studies was medium to high, with scores of > 4 points. Thirty-eight articles (54.3%) scored more than 7 points and 32 articles were rated as medium quality, suggesting that the overall quality of the included studies was acceptable.
The determinants for EGWG
According to the conceptual framework of the SDH, all factors were identified (58 factors, 3 themes—individual [7 aspects, 37 factors]; family [4 aspects, 8 factors]; and social [4 aspects, 13 factors] (Fig. 2). Therefore, a meta-analysis was conducted for 13 factors (including 10 individual factors, 2 family factors, and 1 social factor) (Table 1). The results were statistically concluded as six risk factors and two protective factors for EGWG.
Six studies showed a lower risk of EGWG in older pregnant women [9,10,11,12], but two studies demonstrated the opposite results [13, 14], and the remaining studies did not show a significant association between EGWG and age. The age of 30 years was used as the classification basis according to the most common classification method in the included studies. Meta-analysis showed that the younger age (≤ 30 years) was a risk factor for EGWG (OR, 1.14; 95% CI, 1.10 − 1.19) (Table 1).
According to the division of educational level and the years of education mentioned in the original text, the educational level was divided into high and low levels. Significant heterogeneity was observed in the fixed-effects model (I2 = 93%; P < 0.01). After adjusting to the random-effects model, the results demonstrated that there was no correlation (OR, 1.06; 95% CI, 0.92 − 1.21; P = 0.44) (Table 1).
The employment status of pregnant women was not associated with EGWG [10, 19,20,21,22,23,24,25,26], but the meta-analysis found that unemployed pregnant women were more likely to develop EGWG (OR, 1.07; 95% CI, 1.02 − 1.12) (Table 1).
Physiological and anthropometric factors
Based on the classification methods mentioned in the literature, the prepregnancy body mass index (BMI) was divided into overweight (including obesity), normal weight and underweight. Because the number of pregnant women with obesity was relatively small, the obese group and the overweight groups were combined. Women with prepregnancy overweight (including obesity) were more likely to develop EGWG at normal weight (OR, 2.49; 95% CI, 2.11 − 2.94) and women with prepregnancy underweight had about half the odds of developing EGWG (OR, 0.56; 95% CI, 0.54 − 0.59) (Table 1).
In terms of diet, unreasonable energy intake was a risk factor for EGWG [13, 23, 27,28,29]. A high-calorie diet, such as the Western diet model and high-carbohydrate low-fat intake diet was the risk factor . Higher healthy dietary scores also improved the risk of EGWG . Some dietary patterns were protective factors, such as those based on grains, vegetables, legumes, marine fish, milk, and dairy products and the avoidance of snacks between meals [32, 33]. However, excessive intake of vegetables, fruits, fish, and seafood can increase the risk of EGWG [9, 23, 34].
In terms of exercise, pregnant women who exercised less had a higher chance of EGWG [2, 9, 35]. Regular physical exercise reduced the risk of EGWG in pregnant women [21, 24, 36, 37]. However, one study showed the opposite result .
Smoking status was classified as smoking (including quitting) and nonsmoking to perform a meta-analysis, and the probability of EGWG in pregnant women with smoking was 1.29 times higher than in those without smoking (OR, 1.29; 95% CI, 1.25 − 1.34) (Table 1). Particularly, women exposed to passive smoking in the third trimester had a reduced risk of EGWG .
One study reported that pregnant women who consumed alcohol during pregnancy were at low risk of EGWG , but most studies reported the opposite results [22, 23, 26, 36, 39,40,41], consistent with the findings of the meta-analysis (OR, 1.90; 95% CI, 0.50 − 7.22) (Table 1).
Two studies reported that psychological distress during pregnancy, such as depression  and anxiety , reduced the risk of EGWG. However, other studies reported that psychological distress was not associated with EGWG [24, 36, 43,44,45,46]. Pregnancy stress had no effect on EGWG [23, 36, 38, 44, 45], while depression, stress, anxiety, and pregnancy-related anxiety had positive effects on gestational weight gain, although the specific influence on EGWG was unclear .
Cognition and self-efficacy
Most studies mentioned the relationship between parity and EGWG, 11 of which reported that primipara was more likely to develop EGWG [10, 21, 23, 29, 36, 39, 40, 47,48,49,50]. We divided parity into primiparity and multiparity, and reached the same conclusion after the meta-analysis [OR = 1.30, (95%CI:1.28,1.32)] (Table 1).
The number of antenatal care procedures can affect the weight gain of pregnant women . Pregnant women with adequate antenatal care were at risk for EGWG [39, 41]. However, other studies suggested that the level of antenatal care was not significantly related to EGWG [13, 24, 36, 40, 49, 51]. The findings of our meta-analysis showed that pregnant women with inadequate antenatal care had a low probability of developing EGWG (OR, 0.68; 95% CI, 0.55 − 0.84) (Table 1).
Illness or complications
The presence of gestational diseases or complications also increased the probability of EGWG [10, 36, 48, 51]. Pregnant women with lower limb edema were more likely to have EGWG . However, He et al. found that preeclampsia was not related to EGWG .
Pregnant women with a lower socioeconomic status were reported to more likely to have EGWG [12, 40], but Mariana et al. believed that a higher level of family economic income was a risk factor for EGWG ; however, 11 studies considered that level of family income was not associated with EGWG [2, 20, 22, 24, 29, 30, 41, 50, 51, 53].
Some studies suggested that being unmarried or living alone was a risk factor for EGWG [10, 40, 45]; however, other studies had a different opinion [39, 49, 51]. The findings of our meta-analysis showed that pregnant women who were unmarried or lived alone were more likely to develop EGWG (OR, 1.20; 95% CI, 1.18 − 1.23) (Table 1).
Several studies have mentioned the impact of family support on EGWG, including food security [20, 21, 24, 32]and the housing environment [20, 24]. There was a positive correlation between food security and pregnancy weight gain in one study , while the other studies did not find a significant association [21, 24, 32]. The result of our meta-analysis showed that the OR of EGWG with family support was 1.13 (95% CI, 0.84 − 1.50; P = 0.42) (Table 1).
Race/ethnicity and culture
In America, Caucasians were more likely to develop EGWG than Blacks [39, 47, 49, 54, 55], Latinos [26, 49, 56], and Hispanics [39, 54], but some studies did not show difference between different races and EGWG [21, 36, 43]. Several studies found that race was not associated with EGWG in Canada [40, 50]and Brazil . Another report indicated that Asian women had a higher risk of EGWG than Caucasian women , while a study of Asian populations showed that Malaysian women had a higher risk of EGWG than Chinese women . A Dutch study showed a higher prevalence of EGWG in Europeans . Two studies in China reported that ethnic minority status had no relationship with EGWG , while a study in Belgium showed that EGWG occurred more frequently in ethnic minorities .
One study reported that social support was found to be associated with weight gain during pregnancy , while another reported that there was no significant association . Furthermore, different obstetric institutions or obstetric practitioners did not have a significant relationship with EGWG [19, 26, 50, 51]. Guidance on nutrition for EGWG was discussed in six studies [13, 21, 24, 41, 50, 51], and the results of the meta-analysis showed an OR of 1.08 (95% CI, 0.88 − 1.32) (Table 1).
Several studies discussed the relationship between the payment mode of medical expenses and EGWG during pregnancy [12, 23, 26, 36, 49]. Two studies found that pregnant women without medical insurance were more likely to develop EGWG [11, 23]; however, others suggested that there was no correlation [12, 26, 36].
Persistently low neighborhood deprivation was demonstrated to reduce the risk of developing EGWG . Neighborhood socioeconomic disadvantage had no statistically significant relationship with EGWG .
Analysis of heterogeneity
The consistency of education level and prepregnancy BMI (overweight/obesity) was confirmed by leave‐one‐out sensitivity analysis. However, the combined OR and 95% CI were not significantly affected in any of the study groups, and the difference was not statistically significant. This indicated that the above meta-analysis was stable and reliable (Fig. 3).
For educational level and prepregnancy overweight (including obesity), the above meta‐analysis was also performed within subgroups of studies, defined by the region (i.e., Africa and Asia vs. Europe and Americas), type of study (i.e., cohort study, case–control study, and cross-sectional study), and years (i.e., ≤ 2015 vs. > 2015). The subgroup analysis of the education level, combined with the meta-regression results (P < 0.05), suggested that heterogeneity may be related to different types of studies (Table 2).
Analysis of publication bias
For determinants with more than 10 included studies, a funnel plot was used to evaluate publication bias. The number and distribution of each point on both sides were symmetrical in the funnel plot. Egger’s test results showed that all P values were greater than 0.05, suggesting that there was no publication bias. The results are shown in Additional File 6.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.