The comprehensive systematic search yielded 1470 publications. After excluding duplicated publications (n = 530), the titles and abstracts of the remaining studies were assessed, and unrelated investigations (n = 799) were excluded. Then the full text of 141 relevant studies were completely assessed, and finally, 14 eligible investigations were included. Details regarding the exclusion of certain studies are provided in Supplemental Table 3. The selection approach of articles is shown in Fig. 1.

Flow diagram of search strategy and study selection

A total of 39,399 pregnant women from 14 eligible studies, including 10 cohort [14, 22, 23, 25, 31,32,33,34,35,36], 3 cross-sectional [24, 37, 38], and 1 case–control [39] investigations, were included in this systematic review and meta-analysis. The additional characteristics of the included studies are provided in Table 2. The included studies were published between 2007 and 2022, in the countries of the China [23, 25, 31, 33], United states [14, 22, 35, 36], Germany [32], Turkey [37], Tunisia [39], Brazil [38], the Netherlands [34] and Canada [24]. Nine of the articles were conducted in non-Asian countries [14, 22, 24, 32, 34,35,36, 38, 39], and the remaining studies were from Asian countries [23, 25, 31, 33, 37]. Moreover, 11 publications were from developed countries [14, 22,23,24,25, 31,32,33,34,35,36], and the remaining were conducted in developing countries [37,38,39]. The included studies used different methods for assessment of GDM including, the International Association of the Diabetes and Pregnancy Study Group (IADPSG) [23, 25, 31,32,33, 39], self-reported manner [14], criteria for screening tests for GDM [22, 37], National diabetes data group (NDDG) [24], Dutch midwifery and obstetric guidelines [34], and American diabetes association (ADA) [38]. Additionally, IGT was assessed through the fasting blood test [36] and Oral-glucose-tolerance test (OGTT) [35, 36]. In addition, the dietary intakes were assessed through the use of food frequency questionnaire (FFQ) (n = 9) [14, 22,23,24, 32,33,34,35, 39], dietary recall (n = 3) [25, 31, 38], block rapid food screener (n = 1) [36], and food record (n = 1) [37]. Dietary fat intakes were measured in the periods of pre-pregnancy (n = 2) [14, 32], pregnancy (n = 5) [24, 25, 31, 33, 35, 36, 38], or both periods (n = 1) [23]. All eligible investigations used a random-sampling method for participant selection. Therefore, their populations are representative of the adult population and their results are generalizable to the whole adult population. The statistical analysis of all included were adjusted for both age and body mass index (BMI). Additionally, other important confounders were adjusted, including total energy intake (n = 7) [14, 23, 25, 31, 33, 37, 38], carbohydrate intake (n = 2) [23, 37], protein intake (n = 1) [23], weight gain during pregnancy (n = 7) [23, 25, 31, 33, 36,37,38,39] and history of GDM (n = 2) [22, 25, 38, 39]. As the quality scores of all studies were ≥ 7, the NOS classified all of them as high quality.

Meta-analysis on 8 studies, including 27,831 participants [14, 23,24,25, 31,32,33, 38], indicated that women with the highest total fat consumption (median: