**Vitamin D-Related Risk Factors for Maternal Morbidity during Pregnancy: A Systematic Review**

**Maria M. Morales-Suárez-Varela 1,2, Nazlı Uçar 1, Isabel Peraita-Costa 1,2, María Flores Huertas 1, Jose Miguel Soriano 3, Agustin Llopis-Morales <sup>1</sup> and William B. Grant 4,\***


**Abstract:** Vitamin D has well-defined classical functions related to metabolism and bone health but also has non-classical effects that may influence pregnancy. Maternal morbidity remains a significant health care concern worldwide, despite efforts to improve maternal health. Nutritional deficiencies of vitamin D during pregnancy are related to adverse pregnancy outcomes, but the evidence base is difficult to navigate. The primary purpose of this review is to map the evidence on the effects of deficiencies of vitamin D on pregnancy outcome and the dosage used in such studies. A systematic search was performed for studies on vitamin D status during pregnancy and maternal outcomes. A total of 50 studies came from PubMed, 15 studies came from Cochrane, and 150 studies came from Embase, for a total of 215 articles. After screening, 34 were identified as candidate studies for inclusion. Finally, 28 articles met the inclusion criteria, which originated from 15 countries. The studies included 14 original research studies and 13 review studies conducted between 2012 and 2021. This review was finally limited to the 14 original studies. This systematic review was conducted according to the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines, and the quality and strength of the evidence was evaluated using the Navigation Guide Systematic Review Methodology (SING). We found evidence that supports the idea that supplementary vitamin D for pregnant women is important for reducing the risk of gestational diabetes, hypertension, preeclampsia, early labor, and other complications. The data retrieved from this review are consistent with the hypothesis that adequate vitamin D levels might contribute to a healthy pregnancy.

**Keywords:** gestational diabetes; hypertension; maternal morbidity; preeclampsia; pregnancy; supplementation; vitamin D; 25-hydroxyvitamin D

#### **1. Introduction**

There is evidence of early interest in the relationship between vitamin D status and maternal health outcomes [1].

Vitamin D (D2 or ergocalciferol, D3 or cholecalciferol, or both) is a fat-soluble lipophilic prohormone proven to have many metabolic and biological functions. This vitamin is mainly synthetized in the skin as cholecalciferol through the action of ultraviolet light (vitamin D3), but it is also obtained from diet sources and food supplements such as ergocalciferol (vitamin D2) [2] and food materials such as fish oil, fish flesh, dietary supplements, eggs, butter, fortified foods, liver, and mushrooms. Vitamin D deficiency (serum

**Citation:** Morales-Suárez-Varela, M.M.; Uçar, N.; Peraita-Costa, I.; Huertas, M.F.; Soriano, J.M.; Llopis-Morales, A.; Grant, W.B. Vitamin D-Related Risk Factors for Maternal Morbidity during Pregnancy: A Systematic Review. *Nutrients* **2022**, *14*, 3166. https://doi.org/10.3390/nu14153166

Academic Editor: Marloes Dekker Nitert

Received: 22 June 2022 Accepted: 28 July 2022 Published: 31 July 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

25-hydroxyvitamin D [25(OH)D] < 20 ng/mL) [3–5] is a major public health concern that is widespread among the general population and highly prevalent in pregnant women; it is found in 60% of them [6–9]. Maintaining serum concentrations between 30 and 50 ng/mL is recommended to achieve the health benefits of vitamin D [10–13].

Globally, it has been estimated that a billion people may be affected by vitamin D deficiency or insufficiency [14]. Studies in Ethiopia and India have also found that more than 80% and 60% of pregnant women suffered from vitamin D deficiency, using a cutoff of <50 nmol/L vitamin D, indicating the need for more research on the potential outcome and benefits of supplementation in developing countries [15,16].

Severe maternal morbidity during pregnancy is identified and reported worldwide. Its rising rates remain a large healthcare concern [17]. In 2005, worldwide, there were around 535,900 maternal deaths reported, which translates to a mortality ratio of about 402 maternal deaths per 100,000 live births [18]. The majority of these maternal deaths occurred in sub-Saharan Africa, with 270,500 deaths, and Asia, with 240,600 deaths [18]. Just five countries—India (117,100), Nigeria (58,800), the Democratic Republic of Congo (32,300), Afghanistan (26,000), and Ethiopia (22,200)—accounted for almost half (48%) of all maternal deaths [18].

Maternal morbidity is an unintended outcome of labor and delivery that results in significant short- or long-term consequences to woman's health [19]. Severe maternal morbidity (SMM) affects around an estimated 50,000 women per year in the United States— 0.5–1.3% of pregnancies [19,20]. However, determining the true rates of SMM in the United States and worldwide is difficult because of the lack of standard definitions of such cases as well as the difficulty in identifying cases [21].

During pregnancy, there are significant alterations in phosphate and calcium metabolism owing to calcium accumulating in the fetal skeleton, and the fetus relies exclusively on the maternal supply of vitamin D, which it receives across the placenta, as it is not capable of synthesizing vitamin D on its own for adequate bone mineral formation [22,23]. A low level of vitamin D during the pregnancy and special attention during the early stage of pregnancy produce less bone mineral content in the fetal skeleton. Calcitriol cord blood concentrations tend to be lower than those found in maternal serum [2–13] due to the fact that calcitriol cannot easily cross the placental barrier [24,25], and parathyroid hormone concentrations are low in the fetus [26]. The high levels of phosphorus and calcium concentrations found in serum also contribute to lower fetal calcitriol concentrations because these factors suppress the expression of renal 25OHD-1-α-hydroxylase (CYP27B1) in the fetus [27].

The recommended daily allowance (RDA) of vitamin D for women in the United States aged 19–50 years, including during pregnancy, is established at 600 IU per day [27]. This recommendation was based on the amount of intake necessary to sustain blood levels of vitamin D above 50 nmol/L for a population with minimal sunlight exposure and was developed solely based on outcomes related to bone health [27]. According to the US Institute of Medicine, it is considered that 1000–1600 IU (25–40 g/day) of supplemental vitamin D is necessary during pregnancy to obtain the highest level of vitamin D3 during this period [28]. This recommendation was contentious, as many researchers have argued that insufficiency should be defined at thresholds of 75 nmol/L or even higher, which would require a much higher intake to reach [29,30]. Nevertheless, some studies [31–33] established that the safe and maximal production of vitamin D (at least 32 ng/mL) is achieved with a supplementation of 4000 IU/day until delivery.

Vitamin D can also be referred to as 25-hidroxyvitamin D or calcidiol, and it is transformed into its active form 1,25-dihydroxyvitamin D by CYP27B1 [33]. This enzyme is mainly located in the kidney but is also significantly expressed in the placenta. Pregnancy represents a special physiological situation due to the important role played by the placenta in the metabolism of this vitamin [34]. The placenta is thought to be the major site of vitamin D metabolism in pregnancy. The 1a-hydroxylase, the 24-hydroxylase, the 25-hydroxylase (CYP2R1), the vitamin D binding protein (VDB), and the vitamin D receptor

(VDR) have all been detected either in trophoblast cultures or in freshly obtained placental tissue [35–38]. Undoubtedly, the placenta can metabolize vitamin D, providing active 1,25-(OH)2 vitamin D in vitro. However, it is unclear to what extent placental vitamin D metabolism contributes to maternal vitamin D status in pregnancy.

Numerous functions have been attributed to vitamin D due to the pleiotropic properties of the vitamin D receptor (VDR) [39]. Increasing scientific evidence points to the role of vitamin D in maternal mortality and morbidity, in addition to its implication in several pathologies. Allergic and autoimmune diseases and even cancer implications have also been postulated [40]. The vitamin D deficiency during pregnancy cause maternal and fetal side effects [41], such as increases the risk of preeclampsia, glucose intolerance, gestational diabetes, preterm birth and hypocalcemia crisis in the mother. As poor skeletal development, dysfunction in both the mother and newborn and increase the risk birth of a small child for gestational age (SGA) [42]. Also in the fetus it is related to an inadequate immune system, wheezing and eczema, and respiratory infections in infants [43,44].

An area of study that has garnered significant attention is the role of vitamin D and its effect on pregnancy. There is a lack of evidence from systematic reviews and meta- analyses to evaluate the association between vitamin D during pregnancy and maternal morbidity. Given the high prevalence of low vitamin D level status during pregnancy and the public health importance of clarifying the role of vitamin D during pregnancy in offspring health, a better understanding of the nonclassical functions of vitamin D in preventing adverse health outcomes in high-risk populations is needed. The aim of the present review is to summarize the primary outcome in order to identify a cut-off value for a serum vitamin D concentration that increases the risk of maternal morbidity during pregnancy and to determine the possibility of supplementation to avoid it.

#### **2. Materials and Methods**

This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [45,46]. The quality and strength of the evidence was evaluated using the Navigation Guide Systematic Review Methodology (SING) [47–49]. Systematic review registration PROSPERO (CDR42022343174).

#### *2.1. Question PECO*

The PECO question (P: population; E: exposure; C: comparison; O: outcome) of the study was "Is there more morbidity in pregnant women with low levels of vitamin D compared to those with adequate levels of vitamin D?", in which P is pregnancy women; E is a low intake/level of vitamin D; C is an adequate intake/level of vitamin D; and O is pregnancy morbidity.

#### *2.2. Literature Search*

The goal of the search strategy was to identify studies that reported the associations between serum vitamin D concentrations or the intake of vitamin D from supplementation or diet during pregnancy and its maternal morbidity affects. First, we performed a literature search to identify publications eligible for inclusion in the PubMed and Embase databases. The keywords included "pregnancy" OR "gestation" AND "vitamin D" AND "morbidity." The search was limited to human subjects and English and Spanish language articles published between 2010 and May 2022. A total of 50 studies were recovered from PubMed, 15 were recovered from Cochrane, and 150 were recovered from Embase, for a total of 215. In the first phase, duplicates were removed, and the reference lists of relevant publications were searched for fresh research that fulfilled the inclusion requirements. Following the first literature search, the reviewers examined the titles and abstracts to locate those that fulfilled the selection criteria. These articles were assessed for eligibility, with the first screening of the articles based on the information available in the abstract and results sections of each study. The initial screening identified 34 candidate studies, of which 28 met

the inclusion and exclusion criteria. The PRISMA flowchart (Figure 1) shows the number of articles at each stage of the screening process.

**Figure 1.** Search strategy: PRISMA flowchart.

#### *2.3. Study Inclusion/Exclusion Criteria and Data Extraction*

The types of studies included in this review meet the following criteria: controlled trials, both randomized and nonrandomized; prospective cohorts; case-control studies; and systematic reviews looking at the effects of vitamin D on maternal morbidity. All studies were longitudinal in nature and focused on how vitamin D levels in pregnancy

were related to maternal morbidity. Specific inclusion/exclusion criteria were developed for the selection of studies to be included in this work, and only published works that met all the criteria were included for review. The selection criteria were the following:


After a thorough assessment by all the authors of the candidate studies, 26 were included in this review.

#### *2.4. Data Extraction*

The data for the present review were retrieved from the previous research articles published earlier. The following data were extracted for the present study: (i) Study characteristics: authors, location and year, type of study, and source of data collection; (ii) sample size; (iii) primary outcome; (iv) findings (maternal morbidity & vitamin D level) (Table 1). The relevant data of the reviews were also summarized in a second table, including: (i) factors analyzed; (ii) gestational week when sample was collected; (iii) vitamin D cutoff (blood sample nmol/L); (iv) vitamin D collected (serum or supplementation); (v) average maternal age (Table 2).




**Table**

**1.**

*Cont.*



index; CM, explant conditioned media; FDR, Food & Drug Administration; GDM, gestational diabetes mellitus; GH, gestational hypertension; HP, healthy pregnant; HUVEC, human umbilical vein endothelial cells; IRB, institutional review board; IUGR, intrauterine growth retardation; MUFA, monounsaturated fatty acids; NC, normotensive control; NOS, Newcastle–Ottawa Scale; OR, odds ratio; PE, preeclampsia or preeclamptic; PTH, parathyroid hormone; PUFA, polyunsaturated fatty acids; RFLP, restriction fragment length polymorphism; SFA, saturated fatty acids; UTI, urinary tract infection; UV, ultraviolet; VDR, vitamin D receptor; VEGF, vascular endothelial growth factor.




mellitus; GH, gestational hypertension; PE, preeclampsia; RFLP, restriction fragment length polymorphism; UTI, urinary tract infection; VDR, vitamin D receptor.

#### *2.5. Study Quaity Assessment*

The quality of the studies is assessed using the following tools: The Eight Star Newcastle–Ottawa Scale (NOS) for observational studies (cohorts and case-controls) [47,48] was used to evaluate the methodological quality—specifically, the risk of bias—of the original studies. Assessment with the Newcastle–Ottawa Scale produces a score ranging from 0 to 9, with the overall score based on three sub-scores based on the subject selection (0–4), the comparability of the subject (0–2), and the clinical outcome (0–3). The study assessment was carried out independently by two individuals (NU and IPC), and discrepancies were brought to a third individual (MMSV) if a compromise could not be reached among the two original individuals after discussion.

Further assessment of the quality of the included studies was carried out using the Scottish Intercollegiate Guidelines Network (SIGN) [49]. Using the SIGN ensures that the validity—including key factors such as bias and confounding—of a study is robustly assessed. The SIGN system in based on the principles of evidence-based medicine, an approach that ensures the use of the most up-to-date, reliable, and scientifically solid evidence available in making decisions about a particular situation being studied [64].

The SIGN system establishes levels of evidence and recommendations to describe a given study and its results. The levels of evidence are based on the study design and the methodological quality of individual studies and are scored from best to worst using the numbers 1, 2, 3, and 4. These scores are further ranked using the ++, +, and—signs. The grades of recommendation, rated from best to worst as A, B, C, and D, are based on the strength of the evidence on which the recommendation is based, and they do not reflect the clinical importance of the recommendation.

#### **3. Results**

#### *3.1. Study Characteristics*

Our search approach yielded up 215 studies identified through database searching; a total of 14 original research studies and 13 review studies remained. After consideration, it was decided to include only the 14 original studies in this review.

Considering the SIGN and NOS scores, the 14 original studies could be regarded as good (high) quality. The important methodological features and the general characteristics of all the review studies are summarized in Table 1. The chosen studies were analyzed according to the design, location and year, source of data, sample size, factor, vitamin D level assessment, and major findings. Meanwhile, the vitamin D analysis details and vitamin D cutoff values of the included articles are listed in Table 2.

The studies were published between 2012 and 2021. The original research studies used data from India [57,58,61], Denmark [59], the United States [54,56], Germany [55], Nigeria [53], the Northern Hemisphere [51], Canada [52], Brazil [50], Egypt [60], Sweden [62], and Mexico [63]. The review research studies included data from Brazil, India, the United States, Puerto Rico, Spain, Iran, and Australia [65–77].

All but six observational studies of vitamin D were conducted in high-income country settings, and most populations had either a presumed risk or a high prevalence of deficiency at baseline (Table 1). The dosing approaches and assay methods in the trials varied: one trial contained multiple intervention arms testing the daily dietary intake of Vitamin D, vitamin D supplementation, and the frequency of UV exposure in the first trimester, in the second trimester, and at the time of delivery. One recent trial tested daily 4400 vs. 400 IU D3. In other studies, the relationship between disease risks was evaluated by measuring serum vitamin D levels with different assay methods (Table 2). This trial [65] showed that a significant effect of sufficient vitamin D status (25OHD ≥ 30 ng/mL) was observed in both early and late pregnancy compared with insufficient levels (25OHD < 30 ng/mL) (OR, 0.28; 95% CI, 0.10–0.96).

Vitamin D supplementation appeared to improve maternal vitamin D levels in the two trials for which data were available [65]. In addition, the results of trials by Christine Rohr Thomsen indicate a seasonal variation effect of the risk of gestational hypertension (*p* = 0.01), PE (*p* = 0.001), and early-onset PE (*p* = 0.014) [51,59]. Women with an estimated date of conception in June had the highest risk of preeclampsia, while women with an estimated date of conception in August had the highest risk of gestational hypertension.

Observational studies of vitamin D status during pregnancy and the risk of preeclampsia have not shown consistent associations. Vitamin D levels were lower (*p* < 0.01) in women with PE [50–52,57,58,60,61]. The investigators of a study from the USA [54] observed that vitamin D supplementation initiated in weeks 10–18 of pregnancy did not reduce preeclampsia incidence in the intention-to-treat paradigm. However, vitamin D levels of 30 ng/mL or higher at trial entry and in late pregnancy were associated with a lower risk of preeclampsia (8.08% vs. 8.33%, respectively; relative risk: 0.97; 95% CI, 0.61–1.53). A nested case control study from North Carolina reported that women with vitamin D levels <50 nmol/L had a nearly fourfold greater risk of severe preeclampsia compared with those with levels ≥ 75 nmol/L [78]. In contrast, a nested case-control study in Massachusetts found no statistically significant differences in the risk of pre-eclampsia for women with vitamin D levels < 37.5 nmol/L (AOR 1.35 [0.40, 4.50]) [71]. Another prospective cohort study of pregnancies at a high risk for pre-eclampsia in Canada found no effect of vitamin D during early pregnancy on pre-eclampsia risk [72].

A group of studies relate the vitamin D status with the alteration of different metabolic pathways such as carbon and peptide metabolism. The imbalance of long-chain polyunsaturated fatty acid metabolites produced by a vitamin D deficiency contributes to inflammation and endothelial dysfunction [61]. This deficiency also contributes to a low antimicrobial peptide metabolism [63], resulting in several urinary infections.

#### *3.2. Original Research Studies*

Nandi and colleagues [58] published a cross-sectional study in 2019. The study included 119 pregnant women (69 normotensive controls [NC] and 50 women with PE). The women with PE had lower maternal and cord serum vitamin D levels (*p* < 0.01 for both) than the NC women. A total of 94% of women in the PE group and 76% in the NC group were deficient in maternal vitamin D levels, while for cord vitamin D levels, 98% of women with PE and 85.2% of NC women were deficient. In 2020, this group reported [61] how the imbalance in the long-chain polyunsaturated fatty acid (LCPUFA) metabolites derived from vitamin D deficiency contributes to placental inflammation and endothelial dysfunction in PE.

Rohr Thomsen and colleagues [59] published a cohort study based on data from the Aarhus Birth Cohort (ABC). Of the 50,665 women included, 4285 (8.5%) were diagnosed with a hypertensive disorder of pregnancy, 1999 (3.9%) were diagnosed with PE, and 2386 were diagnosed (4.7%) with gestational hypertension (GH). The hypertensive disorders of pregnancy, including GH, PE, and early-onset PE, increased the risk for women conceiving during spring and early summer, peaking in midsummer, and later decreasing steadily during late summer and fall to reach the nadir by winter. Seasonal variation was found for GH (*p* = 0.01), PE (*p* = 0.001) and early-onset PE (*p* = 0.01). In another prospective comparative study [68], a significant negative correlation was observed between vitamin D and systolic and diastolic blood pressure in the PE group (*p* < 0.05), whereas no significant correlation was observed between vitamin D and systolic/diastolic blood pressure in the control group. The mean vitamin D level was significantly lower in the PE group than that in the control group (9 ± 5 and 14 ± 8 ng/mL, respectively), with a statistically significant *p* < 0.05. A vitamin D level < 5 ng/mL was associated with a 14.58-fold (95% CI; 12.16–17.55) increase in the odds ratio of PE, whereas a vitamin D level of 5–10 ng/mL was associated with an 11.42-fold (95% CI; 8.26–13.6) increase in the odds ratio of PE.

In 2017, Accortt and colleagues [56] found an association between a higher postpartum allostatic load and an index of multisystem physiological wear and tear, operationalizing emergent chronic disease risk and predicting morbidity and vitamin D. Adding vitamin D deficiency to the allostatic load index produced a stronger association with adverse outcome. Brodowski and colleagues [55] assessed the effect of vitamin D supplementation (4400 vs. 400 IU/day) initiated early in pregnancy (10–18 weeks) on the development of PE. When started at weeks 10–18 of pregnancy, vitamin D supplementation did not reduce the incidence of PE. However, vitamin D levels of ≥30 ng/mL at trial entry and in late pregnancy were associated with a lower risk of PE.

Lawal and colleagues [53] showed that no relationship exists between vitamin D deficiency and GDM. That case-control study had 200 pregnant women; the proportion of cases (*n* = 100) and controls (*n* = 100) with vitamin D insufficiency was 62% and 54%, respectively. Lechtermann and colleagues [51] indicated that patients with PE had lower serum levels of vitamin D in response to seasonal changes.

In 2020, Schoenmakers and colleagues [62] found a correlation between a relatively high concentration of 1,2(OH)2D and hypercalcemia in pregnant women during the third trimester. The retrospective and explorative study investigated the prevalence of hypercalcemia in a cohort of 2121 women—1827 screened for hypercalcemia in T3. The prevalence was 1.7% higher than that in the general population.

Olmos-Ortiz and colleagues suggest [64] cardiovascular risk and perinatal infections due to vitamin D3 (calcitriol) deficiency, especially in male-carrying pregnancies due to the lower calcitriol-activating enzyme. The placental calcitriol was significantly elevated in women with urinary tract infections, and it was negatively correlated with blood pressure. Regarding newborns' sex, the calcitriol-activating enzyme showed a higher expression in female-carrying mothers.

The level of evidence is relatively high—2++ or 2+, according to SIGN, which belong to a great level of recommendation: B. The systematic review about the importance of the maintenance of a good level of vitamin D could be used as a recommendation guide in the studied population: pregnant women.

#### **4. Discussion**

Overall, this systematic review suggests that maternal low levels of vitamin D during pregnancy lead to a greater risk of gestational diabetes, preeclampsia, early labor, and other complications. However, due to the variability in numerous elements of the study design (e.g., vitamin D assessment methods, pregnant mobility assessment methods, and the timing of the data collection), it remains a challenge to synthesize the findings. This data suggest that low maternal vitamin D appears to have a negative impact or detrimental impact on the health status of pregnant women, which is an important conclusion that prevents many women from getting adequate nutrition with the adequate support of vitamin D, and it is not possible to use supplementation during the pregnancy period.

Recently, vitamin D has been recognized as interacting with a nuclear receptor in various organs [71–76]. Vitamin D deficiency is associated with increased risks of morbidity and mortality in cardiovascular, malignant, and autoimmune diseases [72,77,78]. In recent years, the interest in the consequences of maternal vitamin D deficiency and its effect on pregnancy has increased. Vitamin D insufficiency is considered common in pregnant women, and deficiencies have been linked to adverse pregnancy outcomes [78–80].

Considering whether prenatal vitamin D deficiency is associated with maternal morbidity seems reasonable. The findings from several studies suggest an increasing prevalence of vitamin D deficiency in pregnancy and its associated adverse outcomes [81–85]. To further understand the role of vitamin D in pregnancy and the seemingly associated adverse outcomes, interventional and observational studies are needed.

Furthermore, a current systematic review described the overall mean prevalence rates of vitamin D deficiency in pregnant women and newborns as 54% and 75%, respectively [86]. In postpartum periods, the prevalence of vitamin D deficiency in women is also high: 63% [86,87]. Although evidence points to the high prevalence of deficiency, there exist strategies to raise maternal vitamin D concentrations, including supplementation, advice for sun exposure (15–20% of the body surface area), and the intake of vitamin D–fortified foods. The vitamin D status during pregnancy varies around the world as a function of maternal sunlight exposure, the degree of skin pigmentation, latitude, lifestyle, BMI, and

the intake of vitamin D supplements. People with darker skin pigmentation and limited sunlight exposure are at the greatest risk for deficiency [88].

Supplement intake can also play an important role in improving vitamin D status among pregnant women. Taking vitamin D-enriched food and supplements can be advised in order to maintain optimum serum levels during pregnancy. The recommendations for vitamin D intake during pregnancy range from 200 to 4000 IU/day worldwide. The current WHO guideline recommends 200 IU/day of vitamin D supplement intake among pregnant women with vitamin D deficiency in order to reduce the risk of PE, a low birth weight, and a preterm birth [89]. The American Pregnancy Association recommends 100 μg/day of vitamin D intake, a considerably larger amount of vitamin D than the recommended intake of 10 μg/day for women [90]. In China, a daily intake of 600 IU is suggested during pregnancy [91]. In the United Kingdom, it is advised to have a maternal vitamin D intake of 400 IU/day. The United Kingdom Health Department provides free vitamin D supplementation to pregnant women and newborn children [92]. Switzerland follows the Institute of Medicine-recommended nutrient intake: 1500–2000 IU/day for women at risk of vitamin D deficiency and 600 IU for women without such risk [93]. In Canada, pregnant women are suggested to take 400–600 IU/day [94]. In Turkey, free supplementation of vitamin D (1200 IU/day) is provided to all women from early pregnancy to 6 months after delivery [95]. A similar approach to vitamin D supplementation (400 IU/day) is followed in New Zealand for pregnant women identified as being at risk of vitamin D deficiency [96]. Meanwhile, for women not at risk, the ministry of health of New Zealand recommends 200 IU/day [97–99].

After many years of study, researchers at the Medical University of South Carolina College of Medicine suggested 4000 IU/day of vitamin D for pregnant women. The findings suggest that, starting at 12–16 weeks of gestation, vitamin D supplementation at a rate of 4000 IU/day is most effective in achieving vitamin D sufficiency in order to attain an optimal nutritional and hormonal vitamin D status throughout pregnancy [88]. A treatment (<37 weeks) goal > 40 ng/mL was associated with a reduction in preterm birth risk [31].

Further, no trials or observational studies specifically regarding vitamin D supplementation/intake and maternal morbidity during pregnancy were identified. Nevertheless, vitamin D requirements are higher among pregnant women, and maintaining optimum serum levels of vitamin D during maternity and for fetus growth is important. Adequate levels of vitamin D seem to be a determinant at the time of implantation and placentation for the development of preeclampsia. There is not a consensus regarding the vitamin D blood concentration value that predisposes women to maternal morbidity; hence, is not easy to recommend a specific supplementation treatment. The present systematic review lacks the experimental data needed to establish a general cutoff value of vitamin D in order to settle how important it could be to improving the maternal diet with vitamin D supplements. Further exploration of vitamin D's role in pregnancy and its potential role in maternal morbidity would be worthwhile, including maternal age and sexual dimorphism.

#### **5. Strengths and Limitations of This Review**

This study has limitations. First, there were limited data on maternal vitamin D supplementation during pregnancy regarding long-term outcomes. Second, the studies included here show significant methodological differences, which problematizes the obtention of a consensus on the evidence currently available on the relationship between vitamin D and maternal morbidity during pregnancy. In addition, we may not have been able to access all publications on the relationship between vitamin D and maternal morbidity during pregnancy because the area of analysis is limited to studies that are published in English and Spanish and that are available through the PubMed, Cochrane, and Embase databases.

#### **6. Conclusions**

Despite the inherent limitations discussed above that limit the ability to draw conclusions across studies, some important findings were noted. Collectively, the studies suggest

that appropriate levels of vitamin D during pregnancy are associated with less mobility during pregnancy. Pregnant women should be counselled to maintain an adequate intake of vitamin D, with suitable nutritional support to adequately control their levels. In this systematic review of the literature, we found evidence relating vitamin D to maternal morbidity-related outcomes. However, well-designed, randomized vitamin D supplementation trials in pregnant women carried out to determine the optimal vitamin D status and dosing and evaluate the potential effectiveness of supplementation with respect to the risk of maternal morbidity are still greatly needed.

**Author Contributions:** Conceptualization: W.B.G. Methodology: N.U., W.B.G., I.P.-C., J.M.S., A.L.-M. and M.M.M.-S.-V.; Writing—Original Draft Preparation: N.U., M.F.H. and M.M.M.-S.-V.; Writing— Review & Editing: N.U., W.B.G., I.P.-C., J.M.S., A.L.-M. and M.M.M.-S.-V. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** WBG receives funding from Bio-Tech Pharmacal Inc. (Fayetteville, AR, USA). The other authors declare no conflict of interest.

#### **References**


## *Article* **Urbanization and Unfavorable Changes in Metabolic Profiles: A Prospective Cohort Study of Indonesian Young Adults**

**Farid Kurniawan 1,2,3,\*, Mikhael D. Manurung 3, Dante S. Harbuwono 1,2, Em Yunir 1,2, Roula Tsonaka 4, Tika Pradnjaparamita 2, Dhanasari Vidiawati 5,6, Angelica Anggunadi 7, Pradana Soewondo 1,2, Maria Yazdanbakhsh 3, Erliyani Sartono <sup>3</sup> and Dicky L. Tahapary 1,2**


**Abstract:** The substantial increase in the prevalence of non-communicable diseases in Indonesia might be driven by rapid socio-economic development through urbanization. Here, we carried out a longitudinal 1-year follow-up study to evaluate the effect of urbanization, an important determinant of health, on metabolic profiles of young Indonesian adults. University freshmen/women in Jakarta, aged 16–25 years, who either had recently migrated from rural areas or originated from urban settings were studied. Anthropometry, dietary intake, and physical activity, as well as fasting blood glucose and insulin, leptin, and adiponectin were measured at baseline and repeated at one year follow-up. At baseline, 106 urban and 83 rural subjects were recruited, of which 81 urban and 66 rural were followed up. At baseline, rural subjects had better adiposity profiles, whole-body insulin resistance, and adipokine levels compared to their urban counterparts. After 1-year, rural subjects experienced an almost twice higher increase in BMI than urban subjects (estimate (95%CI): 1.23 (0.94; 1.52) and 0.69 (0.43; 0.95) for rural and urban subjects, respectively, Pint < 0.01). Fat intake served as the major dietary component, which partially mediates the differences in BMI between urban and rural group at baseline. It also contributed to the changes in BMI over time for both groups, although it does not explain the enhanced gain of BMI in rural subjects. A significantly higher increase of leptin/adiponectin ratio was also seen in rural subjects after 1-year of living in an urban area. In conclusion, urbanization was associated with less favorable changes in adiposity and adipokine profiles in a population of young Indonesian adults.

**Keywords:** urbanization; adiposity; dietary intake; adipokines; young adults; prospective cohort

#### **1. Introduction**

As a low-middle income country, Indonesia is facing two major health problems. On the one hand, an increasing prevalence of non-communicable diseases such as cardiovascular diseases (CVD), obesity, and type 2 diabetes (T2D) is becoming rampant. While on the other, infectious diseases such as helminth infections, malaria, and tuberculosis are still highly prevalent in some rural areas, resulting in stark differences of these disease patterns between urban and rural settings [1,2].

**Citation:** Kurniawan, F.; Manurung, M.D.; Harbuwono, D.S.; Yunir, E.; Tsonaka, R.; Pradnjaparamita, T.; Vidiawati, D.; Anggunadi, A.; Soewondo, P.; Yazdanbakhsh, M.; et al. Urbanization and Unfavorable Changes in Metabolic Profiles: A Prospective Cohort Study of Indonesian Young Adults. *Nutrients* **2022**, *14*, 3326. https://doi.org/ 10.3390/nu14163326

Academic Editors: William B. Grant and Ronan Lordan

Received: 8 July 2022 Accepted: 11 August 2022 Published: 14 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

People residing in urban areas are characterized by relatively high caloric and fat intake compared to their rural counterparts [3]. Moreover, urban people tend to be less physically active [4]. These factors can cause a disruption in energy homeostasis, with a surplus stored in the body as fat [5]. Increasing body fat increases the chance of obesity [6]. Previous studies have shown that higher adipose tissue mass is associated with higher inflammation and insulin resistance [7], which eventually could lead to T2D [8] and CVD [9].

Rapid socio-economic development in Indonesia has promoted the migration of people from rural to urban areas to seek a better life [10]. Previous studies have shown that urbanization is associated with new environmental and lifestyle changes that have the potential to put rural individuals at risk of deteriorating metabolic health [11–13]. The limitation of these previous studies evaluating the effect of urban-rural environment on metabolic health is their cross-sectional design, which lacks the power to show causality.

The worldwide increase of obesity is not only observed in older populations but also in young adults [14]. Based on the Indonesian National Basic Health Survey 2018, there is a high burden of obesity and prediabetes in the young adult population [15]. As this population constitutes a significant proportion of Indonesians [16], the increase in the prevalence of these diseases may become a major health burden.

Early problem identification and intervention targeted towards this economically active young adult population in the context of metabolic health could have a great impact on decreasing the incidence rate, or even lowering the prevalence of non-communicable diseases. To this end, we conducted a prospective cohort study to assess the effect of urbanization over time and its contributing factors on the metabolic health profiles of the Indonesian young adult population.

#### **2. Methods**

#### *2.1. Study Design and Population*

This prospective cohort study was conducted on the Depok campus of the University of Indonesia (UI). Freshmen/women UI bachelor students were recruited in this study. Baseline data were collected in the first three months of the start of the academic year, between August-November 2018, while the follow-up sample collection was performed one year later. Subjects' recruitment was started by providing information about the study during the medical examination of newly arrived students, via social media, and by spreading flyers/leaflets after classes, as well as in student dormitories. A short interview was performed to collect information regarding the areas where the students originated from. Afterwards, a detailed explanation of the study was given to the subjects who agreed to participate and fulfilled the criteria set in this study. After written informed consent, subjects were invited to visit the Makara UI Satellite Clinic to undergo clinical assessment, measurements, and blood sampling. The subjects were classified into the urban group if they were born and lived in urban areas, such as in Jakarta metropolitan areas or in one of the provincial capital cities. The rural group comprised subjects that were originally born and lived in rural areas, defined as the villages that are located at the district levels across Indonesia. Pregnancy and students with previously known diabetes, prediabetes, severe liver or kidney dysfunction, cardiovascular and autoimmune diseases were excluded from the study. This study was approved by the Ethical Committee of Faculty of Medicine Universitas Indonesia (No. 1181/UN2.F1/ETIK/2017).

#### *2.2. Anthropometric Measurements*

Body height was measured using a portable stadiometer (SECA Model 213, Seca Gmbh Co., Hamburg, Germany), while body weight and body composition were measured using a Tanita body impedance analyzer (TBF-300A, Tanita Corp, Tokyo, Japan). Body mass index (BMI) was calculated in kg divided by squared height in meters. Three measurements of waist circumference were taken for each subject using an ergonomic circumference measuring tape (SECA Model 201, Seca Gmbh Co., Hamburg, Germany) and according to the WHO standardized protocol. The average of all three measurements was then used for analysis.

#### *2.3. Fasting Blood Glucose, HbA1c, Fasting Insulin, and HOMA-IR Measurement*

All clinical measurements and blood samples collection were performed after overnight fasting. Finger prick blood was used for measurement of fasting blood glucose (Accu-Check Performa, Roche Diagnostic GmBH, Germany) and HbA1c (A1c EZ 2.0 HbA1c Analyzer, BioHermes, Wuxi, China) levels. The results of fasting blood glucose (FBG) and HbA1c were used to detect subjects with undiagnosed diabetes and prediabetes that had to be excluded from the study. Serum fasting insulin levels were measured in a certified commercial laboratory (Prodia Lab) by a solid-phase, enzyme-labeled chemiluminescent immunometric assay (Siemens IMMULITE 2000XPi) with an assay range of 2–300 mU/L. For the levels below 2 mU/L, a standardized formula from the instrument manufacturer was used to interpolate the concentrations. Homeostatic model assessment for insulin resistance (HOMA-IR) as a validated measure for whole-body insulin resistance (IR) in humans was calculated using the formula: HOMA-IR = fasting serum insulin × fasting glucose/22.5) [17].

#### *2.4. Leptin, Adiponectin, and Leptin/Adiponectin Ratio*

Serum leptin and adiponectin levels were measured by ELISA using commercial reagents (DuoSet ELISA R&D System) according to the manufacturer's protocol. Leptin to adiponectin (L/A) ratio, a more sensitive marker for adipose tissue dysfunction, was calculated by L/A = leptin level (ng/mL)/adiponectin level (μg/mL) [18].

#### *2.5. Dietary Intake Analysis*

One week before the intended measurement date, each subject was informed and instructed on how to make a 3-day food record consisting of two working days and one day during the weekend. For each recording day, all participants were required to write down all of the food and drink they consumed throughout the day. The household servings portion for each meal, food preparation methods, brand name of the foods or beverages if applicable, as well as the addition of sugar, were recorded, as described previously [19]. On the study subjects' clinical measurement and blood sampling day, a certified dietician performed an interview with the subjects to review the completeness and validity of the food record data. These dietary intake data were then analyzed using NutriSurvey 2007 (EBISpro, Willstatt, Germany) software. The amount of total calorie, carbohydrate, fat, and protein intake for each day were obtained and then averaged for further analysis, as published [20].

#### *2.6. Physical Activity Analysis*

Physical activity was assessed using the adapted Global Physical Activity Questionnaire (GPAQ), which was developed by the World Health Organization [21] and validated for the Indonesian population [22]. This self-reported questionnaire comprised 16 questions that were grouped to collect information regarding physical activity over a typical week in three domains: activity at work, transportation (travel to and from places), and recreational activity [21]. All subjects were asked to fill in the questionnaire based on their one-week activities before the measurement date. According to GPAQ analysis guidelines [23], an estimation of the total weekly volume of moderate and vigorous physical activities (MVPA) was given as Metabolic Equivalent-minutes/week (MET. minutes/week), along with the total time spent on MVPA (minutes/week) and total time of sedentary activities in one week (minutes/week) [24]. Furthermore, based on their total volume and time spent on MVPA, the subject's physical activity level was classified into three categories (low, moderate, and high) [23].

#### *2.7. Statistical Analysis*

Continuous variables with normal distribution were presented as mean and standard deviation [mean (SD)]. Meanwhile, non-normally distributed data were presented as geometric mean and 95% confidence interval (geomean (95%CI)) and were log-transformed (log2) for analysis. Linear regression (IBM SPSS Statistics ver. 25) was performed to compare the mean differences of independent variables between two groups at baseline when adjustment for covariates was needed. The chi-square test was used to compare categorical data. Mediation analysis for evaluating the effect of dietary intake components on anthropometry parameter differences between rural and urban group at baseline was performed using PROCESS macro ver. 4.0 for SPSS, as described previously [25].

The changes in parameters measured at baseline and 1-year follow-up for each group, and the differences of these changes between urban and rural subjects, were analyzed using linear-mixed model as implemented in the lme4 R package [26]. For each parameter, the covariates used in the linear mixed model were origin (urban/rural), time, and their interaction. The within subject correlation was accounted for using a random-intercepts term. The statistical significance of the effects (i.e., changes from baseline within each group and between groups) were tested using the F-test with Satterthwaite's degree-of-freedom as implemented in lmerTest [27]. Mediation analysis for the BMI and adipokines changes was performed using 5000 bootstrap samples to obtain the 95% confidence interval for the indirect effect of the covariates. In particular, we evaluated the statistical significance in the decrease/increase of the estimate of the outcome variables after correcting for the changes in certain covariates. Linear mixed model analyses and bootstrapping were performed using R version 4.1.2 in RStudio version 1.4. For all tests, statistical significance was considered at the two-sided 5% level.

#### **3. Results**

#### *3.1. Study Population*

A total of 189 (106 urban; 83 rural) subjects were recruited at baseline. For urban subjects, 87.7% originated from Jakarta metropolitan areas, while the rest were from other provincial capital cities. The overall loss to follow-up was 22.1%, leaving 81 urban and 66 rural subjects at the one-year assessment time point. The main reasons for loss to follow-up were refusal to continue (18 subjects/9.4%), could not be contacted (22 subjects/11.6%), and moved to study at another university (2 subjects/1.1%). The proportion of loss to follow-up was similar between rural and urban groups (see flow-chart of the study in Figure S1).

#### *3.2. Metabolic Profiles of Urban vs. Rural Subjects at Baseline*

Age and proportion of males and females were similar between the rural and urban groups. Adiposity indices (BMI, waist circumference, and fat percentage) were significantly higher in urban compared to rural subjects (mean differences (95%CI) after adjustment for age and sex: 2.81 (1.55; 4.07) kg/m2, *p* < 0.001; 6.37 (3.25; 9.50) cm, *p* < 0.001; and 5.07 (2.70; 7.44) %, *p* < 0.001; for BMI, waist circumference and fat percentage, respectively). Moreover, if BMI was grouped based on the WHO cut-off for Asian populations [28], we observed a higher proportion of overweight/obese in urban compared to rural subjects. Conversely, the proportion of underweight subjects was almost three times higher in the rural than in the urban group (Table 1).


**Table 1.** Baseline characteristics of the study population.

† Not normally distributed continuous variables, presented as geomean (95%CI) and log transformed for analysis. # Analyzed with linear regression for continuous variables and Chi-square test for categorical variables. The *p*-values shown in bold represent the statistically significant differences with *p*<0.05. BMI: body mass index; FBG: fasting blood glucose; HOMA-IR: homeostatic model assessment for insulin resistance.

There was no difference in the fasting blood glucose and HbA1c levels between the two groups. Urban subjects had double the HOMA-IR, leptin levels, and L/A ratio than their rural counterparts. The opposite was observed for adiponectin levels. Further adjustment for BMI revealed that the differences remained significant for L/A ratio, while for HOMA-IR, leptin, and adiponectin became not statistically significant (Table 1).

#### *3.3. Dietary Intake and Physical Activity at Baseline*

Regarding dietary intake, we observed that urban subjects had significantly higher total calorie, fat, and protein intake compared to their rural counterparts (mean differences (95%CI) after adjustment for age and sex: 162.0 (59.4; 264.7) kcal, *p* = 0.002; 8.2 (3.7; 12.6) gram, *p* < 0.001, and 8.4 (4.7; 12.2) gram, *p* < 0.001), for total calorie, fat, and protein intake, respectively) (Table 1). Additionally, the differences in BMI, waist circumference, and fat percentage between the two groups were slightly attenuated after further adjustment for fat and protein intake, despite remaining statistically significant ((2.22 (0.92; 3.52) kg/m2, *p* = 0.001 for BMI; 4.95 (1.74; 8.16) cm, *p* = 0.003 for waist circumference; and 4.38 (1.91; 6.85)%, *p* = 0.001 for fat percentage)). Moreover, mediation analysis showed that fat intake, compared to the other dietary intake components, might be the major driver of the differences in the adiposity profiles between urban and rural subjects at baseline (Table S1).

Next, we compared the physical activity profiles between the two groups at baseline based on the GPAQ analysis. The results showed that urban subjects had higher total volume and total time spent on MVPA compared to their rural counterparts. However, if these parameters were categorized as low, moderate, or high physical activity levels, no statistically significant differences were observed between the two groups. Meanwhile, for the total time of sedentary activities, we observed lower values for urban compared to rural subjects. (Table S2).

#### *3.4. Effect of Urbanization over Time on Adiposity Profiles, Insulin Resistance, and Adipokines*

At follow-up, after one year, both groups experienced an increase in their BMI. When we compared the degree of changes over time, we found that the increase of BMI in rural subjects was almost double what was seen in their urban counterparts (estimate (95%CI) after adjustment for age and sex: 1.23 (0.94; 1.52), *p* < 0.001 and 0.69 (0.43; 0.95), *p* < 0.001, for rural and urban subjects, respectively, Pint < 0.01). Although a similar pattern was observed for fat percentage, the difference between the groups did not reach statistical significance (2.18 (1.39; 2.97), *p* < 0.001 in rural subjects vs. 1.33 (0.62; 2.04), *p* < 0.001 in urban subjects, Pint = 0.12). Meanwhile, HOMA-IR at one-year follow-up did not change significantly compared to baseline in either rural or urban groups (Figure 1).

**Figure 1.** Changes of BMI, fat percentage, and whole-body insulin resistance (HOMA-IR) in urban and rural subjects after 1-year of living in an urban environment. The changes are presented as estimate and 95% confidence interval (95%CI). The changes in each group and the differences of changes between the urban and rural group for each parameter were analyzed using a linear-mixed model, adjusted for age and sex. The *p*-value depicted in the figure represents the *p*-value for interaction (Pint), the level of significance in the differences of changes between the two groups. \* *p* < 0.05. # HOMA-IR was log-transformed (base 2) for analysis. The estimates (95%CI) were back-transformed (2β) and presented as a multiplicative scale compared to baseline. BMI: body mass index; HOMA-IR: homeostatic model assessment for insulin resistance.

Similar analysis was performed for adipokines data, which revealed that both groups had increased leptin levels at 1-year follow-up, with a trend towards a higher increase in rural than urban subjects (Figure 2A, Table 2). Additionally, no changes in the adiponectin levels were observed in urban subjects at the follow-up time point, but a significant decrease was found in the rural subjects (Figure 2B, Table 2). These changes caused no differences in the adiponectin levels between the two groups at 1-year follow-up time point (Table S3). When L/A ratio was considered, a significant three times higher increase was seen in the rural compared to urban group (Figure 2C, Table 2). After further adjustment with the changes in BMI over time, these changes of leptin, adiponectin, and L/A ratio were attenuated and became non-significant for urban subjects (Table 2).

**Figure 2.** Changes of leptin levels (**A**), adiponectin levels (**B**), and leptin-adiponectin (L/A) ratio (**C**) in urban and rural subjects after 1-year of living in an urban environment. The changes are presented as estimate and 95% confidence interval (95%CI). The changes in each group and the differences of changes between urban and rural group for each parameter were analyzed using a linear-mixed model, adjusted for age and sex. All parameters were log-transformed (base 2) for analysis. The estimates (95%CIs) were back-transformed (2β) and presented as percent changes compared to baseline. The *p*-value depicted in the figure represents the *p*-value for interaction (Pint), the level of significance in the differences of changes between the two groups. \* *p* < 0.05.

**Table 2.** Mediation analysis of the effect of changes in BMI overtime on the leptin, adiponectin, and L/A ratio in urban and rural subjects at 1-year follow-up.


† All variables were analyzed using a linear mixed model on log transformed data, presented as estimate and 95% confidence interval. # Indirect effect of BMI on the variables analyzed, obtained by performing bootstrapping with 5000 iterations and presented as its 95% confidence interval. BMI: body mass index; L/A ratio: leptin/adiponectin ratio; Pint: *p*-value for interaction.

#### *3.5. Effect of Urbanization over Time on Dietary Intake and Physical Activity*

The changes over time in two important factors associated with urbanization-related lifestyle, namely, dietary intake and physical activity, were considered next. At the followup time point, a significant increase in total calorie, fat, and protein intake was seen in both groups. However, only the increase in protein consumption was significantly higher in rural than in urban subjects ((7.99 (4.42; 11.56), *p* < 0.001 vs. 14.03 (9.95; 18.10), *p* < 0.001, for urban and rural, respectively, Pint = 0.03) (see Figure 3). These changes also resulted in the loss of differences in the protein intake between the two groups at the 1-year follow-up time point (Table S3). Similar to the findings at baseline, adjustment for the increase in fat intake after one year contributed to the largest attenuation of the BMI increase in both groups (in urban: 29.0% vs. 5.8% vs. 1.4%, and in rural: 19.5% vs. 8.9% vs. 7.3%, for fat, protein, and carbohydrate intake, respectively) (Table 3). Although the increase in protein intake was almost twice as high in the rural group compared to the urban group after 1-year, adjustment for protein intake changes did not attenuate the differences in the increase in BMI between the two groups (Table S4).

**Figure 3.** Changes of calorie-, fat-, protein-, and carbohydrate intake in urban and rural subjects after 1-year of living in an urban environment. The changes are presented as estimate and 95% confidence interval (95%CI). The changes in each group and the differences of changes between urban and rural group for each parameter were analyzed using a linear-mixed model, adjusted for age and sex. The *p*-value depicted in the figure represents the *p*-value for interaction (Pint), the level of significance in the differences of changes between the two groups. \* *p* < 0.05.

**Table 3.** Mediation analysis of the effect of changes in dietary intake and physical activity over time on the changes of BMI at 1-year follow-up in both urban and rural subjects.



**Table 3.** *Cont.*

† All variables as an additional adjustment for age and sex, and all analyses were performed using linear-mixed model. The group of covariates used for model adjustment were shown in bold. †† Proportion of changes in the estimate of the model compared to the model adjusted for age and sex only. # Indirect effect of covariate(s) on BMI, obtained by performing bootstrapping with 5000 iterations and presented as its 95% confidence interval. BMI: body mass index; MVPA: moderate-vigorous physical activity; Pint: *p*-value for interaction.

With respect to physical activity, we found a significant decrease in the total volume of MVPA after one year in the urban group only. However, the difference of changes between the two groups was not statistically significant. A similar pattern was also observed for the total time spent on MVPA. Moreover, there was a significantly higher decrease in total sedentary time after one year in the rural group, with a trend for a decrease in the urban group (Figure S2). Furthermore, addition of the physical activity parameters to the model with fat and protein intake did not significantly further attenuate the estimated changes of BMI in either group (Table 3).

#### **4. Discussion**

Here, we report the first prospective cohort study in an Indonesian young adult population that evaluated the effect of urbanization on metabolic health profiles. Our study showed that rural subjects had overall better adiposity, insulin resistance, and adipokine profiles compared to their urban counterparts. Importantly, we observed a significantly higher increase in BMI and leptin/adiponectin ratios in the rural subjects migrating to an urban area compared to subjects originating from urban areas.

The higher adiposity indices, proportion of overweight/obese, and whole-body insulin resistance in urban compared to rural residents of Indonesia have been reported before [29]. Unhealthy dietary behavior, such as high intake of calories and fat-dense foods associated with urban living, is thought to contribute to the higher adiposity profiles [3]. Indeed, we confirmed this pattern of dietary intake in our study. Although all further adjustments with total calorie, fat, or protein intake attenuated the anthropometric differences between rural and urban groups, our study showed that fat intake contributed the most. Additionally, the longitudinal follow-up to see how urban lifestyle affects metabolic health in those migrating from rural areas compared to urban residents, first confirmed a significant increase of BMI after one-year follow-up in both groups, as seen in previous studies, showing that the majority of freshmen gain weight during their first year of university life [30,31]. The increase of total calorie, fat, and protein intake after one year in both groups might partially explain these changes in BMI. Our study also implicated fat intake changes as the dietary intake component that might be the major contributor for BMI increase after one year in both groups. Significantly, the rural group experienced almost a twice higher increase in BMI than the urban group. Although a significantly higher protein intake was observed in rural compared to urban group at one year follow-up, this could not explain the differences in the BMI increase between the two groups. Interestingly, previous studies have shown that higher meat or meat-products intake, which mostly consist of protein and fat, is associated with more weight gain independent of the total calorie intake [32].

Another factor contributing to adiposity profiles is the level of physical activity [33], as this promotes burning of calories, leading to negative energy balance and subsequently less probability for fat deposition [34]. In our study, at baseline, we found that rural group had lower total volume and time spent on MVPA with a higher sedentary time compared to urban group. This suggests that physical activity does not explain the differences observed in BMI. However, it has to be noted that studies of physical activity in rural and urban areas can be influenced by factors such as the level of education, ethnicity, and tools utilized [35,36]. In our study, the questionnaires used at baseline, which took place when the study subjects had already arrived in urban area, might not truly reflect the subjects' level of physical activity during their residence in rural areas. At the 1-year follow-up time point, we found no significant differences in the changes of total volume and time spent on MVPA between the two groups. As for fat or protein intake, physical activity did not explain the higher gain in BMI seen in the rural group.

The addition of the physical activity parameters into the model with the adjustment of dietary intake also could not explain the higher increase of BMI in rural compared to urban group. This result implies that with similar changes of dietary intake and physical activity within one year, rural subjects experienced bigger changes in BMI than their urban counterparts. Hence, other factors, such as the gut microbiome [37] or epigenetic changes [38], could potentially influence the adiposity changes in the rural population upon migration to urban areas. Other factors that could potentially influence the changes of weight or BMI in our study subjects, as shown in previous studies, are psychological stress [39] and socioeconomic-cultural backgrounds [40,41]. These factors were not evaluated in our study.

The increase of BMI in both groups, if continued for the long term, could potentially cause obesity and induce other metabolic and cardiovascular diseases. In other cases, if the BMI increase does not lead to obesity, the distribution of body fat caused by the weight gain also needs to be considered. Previous studies have shown that Asian populations tend to have higher cardiometabolic risks compared to Caucasian populations with the same levels of BMI, in particular, due to central obesity or visceral adiposity [42,43]. These risks would potentially be higher in the rural group as a substantial number of subjects are underweight. Several studies showed that individuals with previous malnourished condition have an increased risk of obesity later in life, especially if they adopt unhealthy lifestyles [44,45]. Moreover, individuals that have experienced a double burden of malnutrition or undernutrition in early life followed by later overweight/obesity, also pose a substantially enhanced risk of non-communicable diseases (NCDs) [46].

The observed differences in insulin resistance and adipokine levels at baseline between urban and rural groups aligned with the findings from previous studies, showing a higher HOMA-IR and lower adiponectin levels in urban compared to rural population [47,48]. Moreover, after adjustment for BMI, the differences in HOMA-IR, leptin, and adiponectin were no longer statistically significant, indicating a major contribution by BMI. Interestingly, after 1-year living in an urban area, a significant decrease in adiponectin levels, along with a significant increase in L/A ratio, was observed in rural groups compared to the urban group. These changes were attenuated after adjustment for the changes in BMI. This finding shows that rural subjects also experienced worse changes in the adipokine profiles, which was partially mediated by the changes in BMI. It also indicates that there might be other factors than BMI, potentially contributing to the changes in adipokines in the rural subjects after 1-year living in an urban area. As shown from previous studies, gut microbiota has been associated with changes in leptin and adiponectin levels in response to a high-fat diet [49,50].

In our study, although urban subjects had a significant increase in BMI levels after one year, this did not cause changes in adiponectin levels compared to the significant decrease observed in the rural group. These changes even led to the loss of the differences in adiponectin levels between the two groups at 1-year follow-up time point. These findings showed that rural individuals tend to be more vulnerable in their metabolic parameters upon changes in BMI.

Leptin and adiponectin have opposite effects on subclinical inflammation and insulin resistance. Leptin upregulates pro-inflammatory cytokines such as tumor necrosis factor-α and interleukin-6, while adiponectin has anti-inflammatory properties [18]. Adipose tissue dysfunction, marked by higher leptin and lower adiponectin levels, has been associated with insulin resistance and the incidence of T2D [51]. However, in our study, there were no significant changes in insulin resistance in the groups studied after one year of residence in an urban area. The relatively short follow-up period and preserved pancreatic beta-cell function in the young adult population might potentially explain this [52].

The longitudinal study design, inclusion of several metabolic health parameters, and incorporation of dietary intake and physical activity measurements were several strong points of our study. There is only one previous prospective cohort study known to the author that has evaluated the effect of urbanization on CVD risk factors and major NCDs [12]. However, this study did not incorporate dietary intake analysis and measurement of biological metabolic markers, such as insulin resistance index, leptin, and adiponectin. Our study also observed the importance of fat intake contribution in the increase of BMI in both the freshmen urban group and the rural individuals who recently migrated to an urban area. Previous study evaluating this freshmen weight gain only took into account eating behavior changes but did not perform detailed dietary intake analysis [53].

However, the relatively small number of subjects and short duration of follow-up could be considered as limitations in our study. The addition of tools to evaluate the quality of dietary intake, such as the Healthy Eating Index and the utilization of health technology devices like an accelerometer to assess physical activity more objectively, could provide more accurate data in future studies. Additionally, the inclusion of psychological stress assessment and questionnaires or tools to accommodate the evaluation of the socioeconomic and cultural aspects would also result in a more comprehensive data for future research. Moreover, investigation of the gut microbiome, epigenetic changes, as well as immunological factors, might shed more light on the mechanisms that underlie rapid changes in the metabolic profiles upon urbanization.

In conclusion, the findings in our study show that adoption of an urban lifestyle could potentially cause poorer metabolic health changes in rural individuals who migrate to an urban area. Our findings, in part, complement a previous study that showed the rising BMI in residents of increasingly urbanizing rural areas in low-middle income countries is due to an increase in low-quality diet [54]. However, it also indicates that there is a more rapid increase in BMI of subjects arriving from rural areas that could not be explained by either diet or physical activity. Therefore, further studies are needed, as it is important for policymakers to design innovative approaches to prevent this negative effect of urbanization in young adult population, with particular attention to those migrating from rural areas.

**Supplementary Materials:** The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/nu14163326/s1. Figure S1. Flow Chart of The Study. Figure S2. The changes of physical activity levels (A & B) and sedentary time (C) in urban and rural subjects after 1-year of living in an urban environment. Table S1. Mediation analysis of the effect of dietary intake on the differences of adiposity profiles in urban and rural subjects at baseline. Table S2. The levels of physical activity and sedentary time measured with GPAQ in urban and rural subjects at baseline. Table S3. Characteristics of study population at 1-year follow-up time. Table S4. The effect of the differences in protein intake changes on the differences of BMI increase after 1-year between urban and rural subjects.

**Author Contributions:** Conceptualization, F.K. and M.Y.; methodology, F.K., A.A., E.S. and D.L.T.; investigation, F.K., D.L.T., D.V. and T.P.; resources, E.S., M.D.M. and T.P.; formal analysis, F.K. and M.D.M.; funding acquisition, D.L.T., D.S.H., E.Y., M.Y. and P.S.; supervision, M.Y., D.S.H., E.Y. and P.S.; validation, R.T.; writing—original draft, F.K. and E.S.; writing—review and editing, D.L.T., M.Y., M.D.M., A.A., R.T., D.S.H., E.Y. and P.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** The study was supported by the grant from Ministry of Research and Technology Republic of Indonesia (Grant No. NKB-1555/UN2.R3.1/HKP.05.00/2019) and PUTI Universitas Indonesia (Grant No. NKB-762/UN2.RST/HKP.05.02/2020). The doctoral study of F.K. was funded by a scholarship from The Indonesian Endowment Fund for Education (Lembaga Pengelola Dana Pendidikan/LPDP) Ministry of Finance the Republic of Indonesia, Ref S-364/LPDP.3/2019. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki and approved by Ethical Committee for Health Research of Faculty of Medicine Universitas Indonesia (No. 1181/UN2.F1/ETIK/2017).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** The data that support the findings of this study are available from the corresponding author (F.K.) upon reasonable request.

**Acknowledgments:** The authors would like to thank all study participants in this study. Thanks are due to all research assistants and secretaries for their help during the field work. The authors would also like to thank Makara UI Satellite Clinic for providing the space and permission to perform all the study subject's recruitment and measurements there. The authors would like to thank Emma Houlder for proofreading the manuscript.

**Conflicts of Interest:** The authors declare that they have no conflict of interest.

#### **References**

