Studies included in the review

Figure 1 presents a PRISMA diagram summarising the literature search process. We used a combination of keywords to achieve a more robust search (see online supplementary material, Supplemental Fig. S1). Our initial search of four electronic databases yielded 14 476 articles involving pregnant and non-pregnant women. After removing duplicates using automation tools (Covidence; https://covidence.org/), 9870 papers remained for title and/or abstract screening. After screening the full texts by applying the inclusion and exclusion criteria, twelve original articles addressing pregnant women and fifteen original articles involving non-pregnant women were included in the analysis.

Characteristics of the included studies

Table 1 shows the characteristics of the studies included in this systematic review. Among studies involving pregnant women, there were two cross-sectional^{(Reference Abbas, Adam and Rayis23,Reference Shin, Lee and Song40)} , two case–control^{(Reference Dao, Sen and Iyer34,Reference Rahma, Lumbanraja and Lubis39)} , seven prospective cohorts^{(Reference Scholing, Olthof and Jonker21,Reference Mayasari, Hu and Chao24,Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35–Reference Mocking, Savitri and Uiterwaal38,Reference Garcia-Valdes, Campoy and Hayes41)} and one retrospective cohort study^{(Reference Tan, Qi and He22)}. Of the twelve studies, three articles recruited participants in their early pregnancy period (n 3; 25 %)^{(Reference Scholing, Olthof and Jonker21,Reference Abbas, Adam and Rayis23,Reference Mocking, Savitri and Uiterwaal38)} , four articles examined participants in the second or third trimester^{(Reference Dao, Sen and Iyer34,Reference Jones, Zhao and Jiang36,Reference Rahma, Lumbanraja and Lubis39,Reference Garcia-Valdes, Campoy and Hayes41)} , four articles examined the participants throughout the pregnancy^{(Reference Mayasari, Hu and Chao24,Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35,Reference Liabsuetrakul37,Reference Shin, Lee and Song40)} and one article recruited participants at the onset of IDA at any time point during the pregnancy^{(Reference Tan, Qi and He22)}. Pre-pregnancy BMI was used to measure overweight/obesity in pregnant women, except in Abbas et al.’s study, which used early-pregnancy BMI^{(Reference Abbas, Adam and Rayis23)}. Other three studies also reported gestational weight gain^{(Reference Tan, Qi and He22,Reference Jones, Zhao and Jiang36,Reference Garcia-Valdes, Campoy and Hayes41)} as an additional weight indicator during pregnancy. In terms of measured outcomes, each study reported different combinations of markers. To measure Fe status in pregnant women population, concentrations of ferritin were most frequently reported (n 8, 67 %)^{(Reference Scholing, Olthof and Jonker21,Reference Abbas, Adam and Rayis23,Reference Mayasari, Hu and Chao24,Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35,Reference Jones, Zhao and Jiang36,Reference Rahma, Lumbanraja and Lubis39–Reference Garcia-Valdes, Campoy and Hayes41)} , followed by Hb (n 6, 50·0 %)^{(Reference Abbas, Adam and Rayis23,Reference Mayasari, Hu and Chao24,Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35–Reference Mocking, Savitri and Uiterwaal38)} and serum Fe (n 5, 41·7 %)^{(Reference Mayasari, Hu and Chao24,Reference Dao, Sen and Iyer34,Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35,Reference Shin, Lee and Song40,Reference Garcia-Valdes, Campoy and Hayes41)} . Serum transferrin receptor (sTfR) was reported in three studies along with Hb and ferritin^{(Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35,Reference Jones, Zhao and Jiang36,Reference Garcia-Valdes, Campoy and Hayes41)} . Few studies reported TS% (n 3)^{(Reference Mayasari, Hu and Chao24,Reference Dao, Sen and Iyer34,Reference Garcia-Valdes, Campoy and Hayes41)} . In terms of inflammatory markers, CRP level was the most commonly used marker, which was employed in three studies^{(Reference Dao, Sen and Iyer34–Reference Jones, Zhao and Jiang36)}, followed by IL-6 in two studies^{(Reference Dao, Sen and Iyer34,Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35)} and leptin in one study^{(Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35)}. Flores-Quijano et al. measured IL-6 and leptin apart from CRP levels^{(Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35)}. Hepcidin was only measured by five out of twelve reviewed studies^{(Reference Mayasari, Hu and Chao24,Reference Dao, Sen and Iyer34,Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35,Reference Rahma, Lumbanraja and Lubis39,Reference Garcia-Valdes, Campoy and Hayes41)} . Almost all studies (n 11, 92 %) compared the overweight/obesity group with the non-overweight/obesity group in the analysis^{(Reference Scholing, Olthof and Jonker21–Reference Mayasari, Hu and Chao24,Reference Dao, Sen and Iyer34–Reference Liabsuetrakul37,Reference Rahma, Lumbanraja and Lubis39–Reference Garcia-Valdes, Campoy and Hayes41)} ; only one study did not report a non-obese control group and rather categorised the BMI into tertiles^{(Reference Mocking, Savitri and Uiterwaal38)}. Based on the methodological quality assessment, the overall study quality was moderate. Most studies (n 10) met at least four out of six quality indicators (see online supplementary material, Supplementary Table 1). Scores ranged from four to six, with a mean score of 4·7. One study that did not sufficiently describe the statistical analysis as a quality assessment indicator was excluded from the analysis.

Table 1

Characteristics of the included studies

AGP, alpha(1)-acid glycoprotein; OWT, overweight; OB, obesity/obese; NW, normal weight; CRP, C-reactive protein; Lp, leptin; MCH, mean corpuscular haemoglobin; MCV, mean corpuscular volume; MCHC, mean corpuscular haemoglobin concentration; sTfR, soluble transferrin receptor; SF, serum ferritin; TS%, transferrin saturation; TIBC, total iron-binding capacity; IDA, iron deficiency anaemia; HH, household; DHS, Demographic Health Survey; MUAC, mid-upper arm circumference; NPNL, non-pregnant non-lactating; VLED, very low energy diet; DMPA, depot medroxyprogesterone acetate; IUD, intra-uterine device; OCP, oral contraceptive pill.

We also found that four studies (33 %) did not consider any confounders, such as maternal characteristics, dietary intake, supplement use or the presence of infectious diseases, in their analysis, which might have affected the reported results^{(Reference Abbas, Adam and Rayis23,Reference Dao, Sen and Iyer34,Reference Rahma, Lumbanraja and Lubis39,Reference Garcia-Valdes, Campoy and Hayes41)} . In the studies that considered the effect of confounding variables in the statistical analysis, the most frequently considered confounders were maternal age, gestational age and socio-economic status^{(Reference Scholing, Olthof and Jonker21,Reference Tan, Qi and He22,Reference Mayasari, Hu and Chao24,Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35–Reference Mocking, Savitri and Uiterwaal38,Reference Shin, Lee and Song40)} . Tan et al.’s study reported the most comprehensive list of identified confounders, including maternal age, race, education, local citizens, area of residence, annual family income, multiple gestations, parity, gestational age at the time of the survey, egg intake per week, meat intake per week, smoking before pregnancy, nausea and/or vomiting during pregnancy, multivitamin supplementation, Ca supplementation and multiple gestational co-morbidities^{(Reference Tan, Qi and He22)}. However, their study failed to include Fe intake, birth spacing, infections and inflammatory disorders. Despite Fe supplementation being a well-known confounder, three of the twelve reviewed studies controlled this variable^{(Reference Tan, Qi and He22,Reference Mayasari, Hu and Chao24,Reference Jones, Zhao and Jiang36)} .

In non-pregnant populations, fifteen studies were included^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Cheng, Bryant and Rooney25–Reference Kordas, Centeno and Pachón28,Reference Cepeda-Lopez, Aeberli and Zimmermann30,Reference Rad, Sefidgar and Tamadoni42–Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50)} . The majority of the studies used cross-sectional designs (n 13, 86·7 %)^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Cheng, Bryant and Rooney25,Reference Jordaan, Van den Berg and Van Rooyen26,Reference Kordas, Centeno and Pachón28,Reference Rad, Sefidgar and Tamadoni42,Reference Chang, Chen and Owaga44–Reference Cepeda-Lopez, Osendarp and Melse-Boonstra51)} . Among the remaining two studies, one was an interventional study (43) and the other was a cohort study^{(Reference Karl, Lieberman and Cable27)}. In terms of Fe markers, Hb level was the most frequently reported marker used in non-pregnant women (n 12; 80 %)^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Cheng, Bryant and Rooney25–Reference Kordas, Centeno and Pachón28,Reference Rad, Sefidgar and Tamadoni42–Reference Chang, Chen and Owaga44,Reference Fricker, Le Moel and Apfelbaum46,Reference Hiremath, Kumar and Huchchannavar47,Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50,Reference Cepeda-Lopez, Osendarp and Melse-Boonstra51)} , followed by serum ferritin (n 9, 60·0 %)^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Cheng, Bryant and Rooney25–Reference Kordas, Centeno and Pachón28,Reference Beard, Borel and Peterson43,Reference Chang, Chen and Owaga44,Reference Fricker, Le Moel and Apfelbaum46,Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50)} , TS% (n 6, 40 %)^{(Reference Cheng, Bryant and Rooney25–Reference Karl, Lieberman and Cable27,Reference Beard, Borel and Peterson43,Reference Fricker, Le Moel and Apfelbaum46,Reference Cepeda-Lopez, Osendarp and Melse-Boonstra51)} , serum/plasma Fe (n 5, 33·3 %)^{(Reference Cheng, Bryant and Rooney25,Reference Beard, Borel and Peterson43,Reference Chang, Chen and Owaga44,Reference Fricker, Le Moel and Apfelbaum46,Reference Cepeda-Lopez, Osendarp and Melse-Boonstra51)} and sTfR (n 4, 26·7 %)^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Cheng, Bryant and Rooney25,Reference Karl, Lieberman and Cable27,Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50)} . For inflammatory markers, most studies used CRP level (n 6, 40 %)^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Cheng, Bryant and Rooney25,Reference Jordaan, Van den Berg and Van Rooyen26,Reference Kordas, Centeno and Pachón28,Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50,Reference Cepeda-Lopez, Osendarp and Melse-Boonstra51)} , followed by AGP (n 1, 6·7 %)^{(Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50)} and TNF-α (TNF-α) (n 1, 6·7 %)^{(Reference Karl, Lieberman and Cable27)}. Hepcidin level was only measured by two out of fifteen studies (13·3 %)^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Cheng, Bryant and Rooney25)} . Among the studies conducted on non-pregnant women, only 73 % of the studies met at least four quality assessment indicators. In addition, 40 % of the studies did not consider the effects of any potential confounders in their statistical analyses, which might have introduced significant bias while concluding the findings^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Garcia-Valdes, Campoy and Hayes41,Reference Rad, Sefidgar and Tamadoni42,Reference Chang, Chen and Owaga44,Reference Hiremath, Kumar and Huchchannavar47,Reference Hisa, Haruna and Hikita48)} . Studies conducted by Fricker et al. ^{(Reference Fricker, Le Moel and Apfelbaum46)} and Nainggolan et al. ^{(Reference Nainggolan, Hapsari and Titaley49)} measured sex, menopausal status, oral contraceptive use, blood donation, Fe treatment/drugs, amenorrhoea, chronic disease, education level, physical activity, consumption of fruits and vegetables, and the presence of infectious or non-communicable diseases as confounders; however, both failed to account for dietary intake (Fe intake, Fe enhancers and inhibitors) and supplement use. Overall, quality scores ranged from 2 to 6, with a mean score of 4·1. Studies with a quality score of less than 2 usually did not report outcome measurements, statistical analysis or sampling procedures, and those studies were excluded from the analysis (see online supplementary material, Supplementary Table 2).

Outcomes in pregnant women participants

Summaries of the extracted data for the pregnant women are presented in see online supplementary material, Supplementary Table 3. The prevalence of overweight/obesity among pregnant women in this review ranged from 10·8 % to 69·8 %^{(Reference Scholing, Olthof and Jonker21–Reference Mayasari, Hu and Chao24,Reference Jones, Zhao and Jiang36,Reference Liabsuetrakul37,Reference Shin, Lee and Song40)} . In comparison, anaemia prevalence ranged from 12·7 % to 57·7 %^{(Reference Abbas, Adam and Rayis23,Reference Liabsuetrakul37)} . Various Fe markers were reported in the reviewed studies, as previously mentioned. Out of six studies that measured Hb, three reported a significant positive association between overweight/obesity and Hb levels,^{(Reference Mayasari, Hu and Chao24,Reference Jones, Zhao and Jiang36,Reference Mocking, Savitri and Uiterwaal38)} while the others found no association. Three (out of five) reported a decrease in serum Fe^{(Reference Scholing, Olthof and Jonker21,Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35,Reference Shin, Lee and Song40)} among overweight/obese pregnant women, while the other studies demonstrated no association.^{(Reference Dao, Sen and Iyer34,Reference Garcia-Valdes, Campoy and Hayes41)} . One out of three studies showed a decrease in TS%^{(Reference Mayasari, Hu and Chao24)}, while the others (n 2) showed no association^{(Reference Dao, Sen and Iyer34,Reference Garcia-Valdes, Campoy and Hayes41)} . One study (out of seven) revealed a decrease in serum ferritin^{(Reference Abbas, Adam and Rayis23)} among overweight/obese individuals, one showed an increase in serum ferritin^{(Reference Mayasari, Hu and Chao24)} and the other five studies did not report any association^{(Reference Scholing, Olthof and Jonker21,Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35,Reference Jones, Zhao and Jiang36,Reference Rahma, Lumbanraja and Lubis39,Reference Shin, Lee and Song40)} . Among the seven studies that measured serum ferritin, only two measured inflammatory markers^{(Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35,Reference Jones, Zhao and Jiang36)} . However, they did not adjust serum ferritin levels for the presence of inflammation. Lastly, one (out of three) study reported a significant increase in sTfR levels,^{(Reference Jones, Zhao and Jiang36)} while the other two did not report any difference^{(Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35,Reference Garcia-Valdes, Campoy and Hayes41)} .

Only three studies measured CRP, and all of them reported an increased level in overweight/obese pregnant women^{(Reference Scholing, Olthof and Jonker21,Reference Dao, Sen and Iyer34,Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35)} . An increase in leptin and IL-6 levels was also found by Flores-Quijano et al. ^{(Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35)}. However, one study reported that there was no difference in IL-6 levels between overweight/obese and normal weight^{(Reference Dao, Sen and Iyer34)}. Moreover, four out of five studies found significantly higher hepcidin in overweight/obese pregnant women compared to normal weight^{(Reference Mayasari, Hu and Chao24,Reference Dao, Sen and Iyer34,Reference Flores-Quijano, Vega-Sánchez and Tolentino-Dolores35,Reference Garcia-Valdes, Campoy and Hayes41)} , while another one failed to find a significant association^{(Reference Rahma, Lumbanraja and Lubis39)}. In pregnant women, overweight/obesity was found to be associated with a significantly higher proportion of ID in two studies^{(Reference Abbas, Adam and Rayis23,Reference Mayasari, Hu and Chao24)} . Despite this, out of the five studies that compared the proportion of anaemia between overweight/obesity and normal weight, four reported a lower proportion of anaemia among overweight/obese women^{(Reference Abbas, Adam and Rayis23,Reference Mayasari, Hu and Chao24,Reference Liabsuetrakul37,Reference Mocking, Savitri and Uiterwaal38)} , while one reported no difference^{(Reference Jones, Zhao and Jiang36)}. Among three studies that measured IDA, two found a lower prevalence of IDA among overweight/obese women^{(Reference Tan, Qi and He22,Reference Mayasari, Hu and Chao24)} , while the other found no association between overweight/obesity and IDA^{(Reference Abbas, Adam and Rayis23)}.

A retrospective cohort study in China found that underweight women had a higher risk of IDA compared to normal-weight women. Although the risk was lower in women with obesity, the higher rate of gestational weight gain was associated with a higher risk of IDA^{(Reference Tan, Qi and He22)}. Abbas et al. ^{(Reference Abbas, Adam and Rayis23)} in their study found a relationship between obesity and ID among pregnant women, where obese pregnant women had significantly lower serum ferritin levels (P = 0·010), and thus, had a higher proportion of ID (P = 0·015) than their counterparts. Surprisingly, the same study also reported a lower prevalence of anaemia (P < 0·001) in overweight/obese women without any significant difference in Hb concentration. In conclusion, overweight/obesity in pregnant women studies were generally associated with higher levels of Hb, hepcidin and inflammatory markers, and lower serum Fe, but not associated with the serum ferritin, serum TS% and sTfR. This resulted in a lower proportion of anaemia and IDA in overweight/obese pregnant women.

Outcomes in non-pregnant women participants

The extracted data for non-pregnant women are summarised in see online supplementary material, Supplementary Table 4. Prevalence of overweight/obesity ranged from 25·3 % to 70·78 %^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Karl, Lieberman and Cable27,Reference Kordas, Centeno and Pachón28,Reference Chang, Chen and Owaga44,Reference Eckhardt, Torheim and Monterrubio45,Reference Hiremath, Kumar and Huchchannavar47,Reference Hisa, Haruna and Hikita48,Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50,Reference Cepeda-Lopez, Osendarp and Melse-Boonstra51)} . Anaemia prevalence in non-pregnant women ranged from 10·0 % to 22·3 %^{(Reference Cheng, Bryant and Rooney25,Reference Kordas, Centeno and Pachón28,Reference Chang, Chen and Owaga44,Reference Hisa, Haruna and Hikita48,Reference Nainggolan, Hapsari and Titaley49)} . Many studies used Hb as a proxy parameter to measure Fe status, and none of them demonstrated a significant decrease in Hb concentration among overweight/obese non-pregnant women. Instead, three studies reported inverse results, in which Hb levels increased among overweight/obese compared to normal-weight non-pregnant women^{(Reference Cheng, Bryant and Rooney25,Reference Kordas, Centeno and Pachón28,Reference Fricker, Le Moel and Apfelbaum46)} . Seven out of twelve studies that measured Hb did not show any association between Hb and overweight/obesity^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Cheng, Bryant and Rooney25–Reference Karl, Lieberman and Cable27,Reference Rad, Sefidgar and Tamadoni42,Reference Hiremath, Kumar and Huchchannavar47,Reference Cepeda-Lopez, Osendarp and Melse-Boonstra51)} , three found an increase of Hb among overweight/obese women^{(Reference Kordas, Centeno and Pachón28,Reference Chang, Chen and Owaga44,Reference Fricker, Le Moel and Apfelbaum46)} and the rest (n 2) did not test the association^{(Reference Hisa, Haruna and Hikita48,Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50)} . Out of nine studies that measured ferritin^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Cheng, Bryant and Rooney25–Reference Kordas, Centeno and Pachón28,Reference Beard, Borel and Peterson43,Reference Chang, Chen and Owaga44,Reference Fricker, Le Moel and Apfelbaum46,Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50)} , six studies revealed a positive association between overweight/obesity and serum ferritin levels^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Cheng, Bryant and Rooney25,Reference Kordas, Centeno and Pachón28,Reference Beard, Borel and Peterson43,Reference Chang, Chen and Owaga44,Reference Fricker, Le Moel and Apfelbaum46)} . Among those six studies, three did not report any inflammatory marker^{(Reference Beard, Borel and Peterson43,Reference Chang, Chen and Owaga44,Reference Fricker, Le Moel and Apfelbaum46)} , while others that measured inflammatory markers did not adjust ferritin for the presence of inflammation^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Cheng, Bryant and Rooney25,Reference Kordas, Centeno and Pachón28)} . Only one study adjusted ferritin levels for inflammation status, which reported lower ferritin was associated with central adiposity as indicated by waist circumference (OR = 0·59 (0·38–0·91)) but not with BMI^{(Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50)}.

A few other studies have used serum Fe (n 5), TS% (n 6) and sTfR (n 4) to measure Fe status. One study reported a decrease in serum Fe concentration in overweight/obese subjects^{(Reference Cheng, Bryant and Rooney25)}, while two reported the opposite^{(Reference Beard, Borel and Peterson43,Reference Cepeda-Lopez, Osendarp and Melse-Boonstra51)} and two did not find any association^{(Reference Chang, Chen and Owaga44,Reference Fricker, Le Moel and Apfelbaum46)} . Higher serum TS% in overweight/obese non-pregnant women was only seen in one study^{(Reference Beard, Borel and Peterson43)} out of six, whereas two reported lower serum TS%^{(Reference Cheng, Bryant and Rooney25,Reference Jordaan, Van den Berg and Van Rooyen26)} . The rest of the studies (n 3) did not find any difference^{(Reference Karl, Lieberman and Cable27,Reference Fricker, Le Moel and Apfelbaum46,Reference Cepeda-Lopez, Osendarp and Melse-Boonstra51)} . Of the four studies that measured sTfR, one found an increase in sTfR among overweight/obese women^{(Reference Herter-Aeberli, Thankachan and Bose18)}, two showed no association^{(Reference Cheng, Bryant and Rooney25,Reference Karl, Lieberman and Cable27)} and one did not compare sTfR between the groups^{(Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50)}. Herter et al. in their study reported a higher hepcidin level in overweight/obese non-pregnant women^{(Reference Herter-Aeberli, Thankachan and Bose18)}, while the other study reported no association^{(Reference Cheng, Bryant and Rooney25)}. Higher CRP was reported among overweight/obese subjects in all six studies^{(Reference Herter-Aeberli, Thankachan and Bose18,Reference Cheng, Bryant and Rooney25,Reference Jordaan, Van den Berg and Van Rooyen26,Reference Kordas, Centeno and Pachón28,Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50,Reference Cepeda-Lopez, Osendarp and Melse-Boonstra51)} . Six out of seven studies assessing anaemia reported a lower proportion of anaemia among obese women^{(Reference Kordas, Centeno and Pachón28,Reference Chang, Chen and Owaga44,Reference Eckhardt, Torheim and Monterrubio45,Reference Hiremath, Kumar and Huchchannavar47–Reference Nainggolan, Hapsari and Titaley49)} , while the other reported no difference in the anaemia proportion^{(Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50)}. ID was measured in four studies, two of which reported a lower prevalence of ID among overweight/obese women^{(Reference Karl, Lieberman and Cable27,Reference Chang, Chen and Owaga44)} , one of which reported a higher prevalence^{(Reference Cepeda-Lopez, Osendarp and Melse-Boonstra51)} and one of which found no association^{(Reference Kordas, Centeno and Pachón28)}. IDA was reported in only one study in which a lower prevalence of IDA was found among obese women^{(Reference Chang, Chen and Owaga44)}.

In a study among non-pregnant Cuban, anaemia was associated with ID (OR: 3·02 (1·82–5·03)) and erythropoietic deficiency (sTfr >8·3 µg/ml) (OR: 5·62 (3·03–10·39)), but not with overweight (OR:0·80 (0·57–1·12)), central adiposity (OR:0·80 (0·57–1·12)) and inflammation (OR: 1·00 (0·65–1·54))^{(Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50)}. The study also reported that overweight/obesity was correlated with elevated CRP levels (OR: 3·06 (1·89–4·94)) rather than with AGP levels (OR: 1·80 (1·05–3·08))^{(Reference Pita-Rodríguez, Basabe-Tuero and Díaz-Sánchez50)}. In conclusion, studies among non-pregnant women revealed that overweight/obesity was generally associated with higher ferritin and serum Fe levels but not with Hb, sTfR, TS% and total iron-binding capacity concentrations. An increased level of CRP, AGP and hepcidin among obese non-pregnant women were also reported in most of the studies. Overweight/obese non-pregnant women also had a lower prevalence of anaemia and IDA, but the association with ID was mixed.