Background
Cardiovascular diseases (CVD) are primary causes of mortality and morbidity among women worldwide [
1,
2]. The global level of all-age CVD-cause disability-adjusted life-years (DALYs) in 2017 was 156 million the highest fraction among non-communicable diseases DALYs for women globally [
3]. In 2019, an estimated 523 million CVD cases and 18.6 million CVD-related deaths were reported, with a projected increase to more than 23.6 million deaths by 2030 [
4]. Of these deaths, 6.1 million deaths prematurely occurred in people between the ages of 30–70 years due to CVD [
5]. Additionally, there were approximately 275 million women with CVD and 8.9 million (47.8%) of all mortality cases among women globally are also attributable to CVD [
6].
Elevated blood pressure (BP), dyslipidemia and elevated blood glucose, unhealthy diet, overweight and obesity, physical inactivity, tobacco use, and harmful use of alcohol are well-established modifiable risk factors of CVD [
7] but ageing, gender, family history of CVD, ancestry, and socioeconomic status are identified as relevant non-modifiable CVD risk factors [
8]. However, the dynamics and burden of these risks factors are often under-reported among migrants populations, particularly in comparison with populations with similar ancestry. For example, the burden and risk factors of CVD among Filipino Women (FW) in Korea compared to native-born Korean Women (KW) is not well understood.
Migrants from different cultural environments often find it hard to align with the cultural and traditional norms of the host country [
9]. Specific health disparities and susceptibilities are prevalent in each typology and phase of the migration process [
10]. Transition to improved social and economic conditions, and quality health care services has come with complexities associated with changes in dietary habits and nutrition transitions that could substantially modify the CVD risk, incidence, and severity [
11].
A study conducted by the National Institutes of Health on CVD risk factors among Asian Americans in the US revealed that migrants have a lower prevalence of CVD risk factors than the host population [
12]. However, in this community-based longitudinal study in the US, the migrant population seems to have worse CVD outcomes compared with the host population taking into account gender-specific. For example, Filipino American women presented a greater prevalence of diabetes and metabolic syndrome compared with women of Caucasian ancestry [
13]. In addition, Filipino American women were found to have higher rates of heart disease, stroke, coronary heart disease, and hypertension than Hispanic, Non-Hispanic Whites, East Asians, and Asia–Pacific Islanders, and other Asians [
14]. CVD susceptibility among migrant FW in the US has been documented in the literature [
14]. However, the CVD risk factors or susceptibility to CVD among Filipino and Korean women is not understood clearly yet.
Moreover, whether FW migrants in South Korea are susceptible to higher CVD risk factors than Korean women has not been reported. Understanding the CVD risk susceptibilities among migrant populations deserves attention and is required to design context-specific health promotion guidelines and policies to promote migrant health. In particular, the difference in disease prevalence of migrants in the country of origin and those in the host country may provide insights and perspectives that can help researchers and policymakers to understand the etiology of the disease and design the appropriate strategies.
This study assessed and compared the CVD risk factors among FW migrants in Korea, FW in the Philippines, and KW.
Methods
Study population, data sources, and matching
The current analysis included 504 migrant FW in Korea (enrolled in the Filipino Women's Health and Diet Study—FiLWHEL between 2014 and 2016) matched for age (on a ratio of 1:1, age was randomly selected from three datasets and stratified into 20–34, 35–39, 40–44 and 45–57 yielding the same no. of samples for each age group, Table
1. From Tables
3,
4,
5 and
6, the no. of samples varies after the missing data were excluded) with 504 FW in the Philippines (from the Philippine 8th National Nutrition Survey—NNS enrolled in 2013) and 504 KW (from the 6th Korean National Health and Nutrition Examination Survey—KNHANES, enrolled from 2013 to 2015). We randomly selected and matched participants with matching characteristics and information on blood lipids and glucose profiles for statistical analysis in this study. Details of the participants' selection have been reported in Additional file
1: Fig. S1. These three studies used structured questionnaires through on-site face-to-face interviews to collect demographic, socioeconomic, and health-related behaviors. Anthropometry was measured using a non-stretchable tape measure and bioelectric impedance analysis machine. In this analysis, we only use the one-day 24-h recall from three datasets for dietary assessment. It was conducted through an in-person or telephone interview; FiLWHEL, NNS and KNHANES replicate, if not entirely, the USDA five-step multiple-pass methods [
15]. BP was measured while the participant was seated calmly using a sphygmomanometer. Details of the protocols, study designs, recruitment methods, data collection procedures and validation of these studies have been described elsewhere [
16‐
18]. All three studies were ethically conducted following with the principles of the Declaration of Helsinki [
19].
Table 1
Demographic and anthropometric characteristics of FiLWHEL, NNS, and KNHANES women
No. of participantsd | 504 | 504 | 504 | |
Age, years | 34.46 ± 8.10 | 34.47 ± 8.09 | 34.01 ± 8.05 | matched |
Age (years), % | | | | matched |
20–34 | 277 (54.96) | 277 (54.96) | 277 (54.96) | |
35–39 | 99 (19.64) | 99 (19.64) | 99 (19.64) | |
40–44 | 68 (13.49) | 68 (13.49) | 68 (13.49) | |
45–57 | 60 (11.90) | 60 (11.90) | 60 (11.90) | |
Height, cm | 153.68 ± 5.38 | 151.64 ± 5.40 | 160.52 ± 5.76 | < .0001 |
Weight, kg | 55.81 ± 9.82 | 54.46 ± 10.81 | 57.16 ± 9.52 | < .0001 |
BMI, kg/m2 | 23.62 ± 3.81 | 23.63 ± 4.36 | 22.18 ± 3.47 | < .0001 |
BMI (kg/m2), % | | | | < .0001 |
< 18.5 | 22 (4.42) | 46 (9.18) | 57 (11.31) | |
18.5– < 23.0 | 215 (43.17) | 206 (41.12) | 276 (54.76) | |
23.0– < 25.0 | 107 (21.49) | 81 (16.17) | 72 (14.29) | |
25.0– < 27.0 | 70 (14.06) | 70 (13.97) | 50 (9.92) | |
27.0– < 30.0 | 55 (11.04) | 59 (11.78) | 32 (6.35) | |
≥ 30.0 | 29 (5.82) | 39 (7.78) | 17 (3.37) | |
Waist circumference, cm | 79.74 ± 9.45 | 78.36 ± 11.21 | 75.16 ± 8.90 | < .0001 |
Waist circumference (cm), % | | | | < .0001 |
< 80.0 | 274 (55.35) | 276 (58.60) | 376 (74.60) | |
80.0– < 85.0 | 91 (18.38) | 75 (15.92) | 55 (10.91) | |
85.0– < 90.0 | 66 (13.33) | 51 (10.83) | 36 (7.14) | |
≥ 90.0 | 64 (12.93) | 69 (14.65) | 37 (7.34) | |
Education, % | | | | < .0001 |
High school or below | 164 (32.80) | 342 (68.54) | 170 (36.17) | |
College or above | 336 (67.20) | 157 (31.46) | 300 (63.83) | |
Occupation, % | | | | < .0001 |
Unemployed | 261 (52.10) | 345 (68.45) | 185 (39.53) | |
Employed | 240 (47.90) | 159 (31.55) | 283 (60.47) | |
Smoking status, % | | | | 0.0031 |
Never smoker | 454 (91.90) | 417 (87.24) | 413 (84.29) | |
Past smoker | 38 (7.63) | 43 (9.00) | 50 (10.20) | |
Current smoker | 6 (1.20) | 18 (3.77) | 27 (5.51) | |
Alcohol intake, % | | | | < .0001 |
Never drinker | 148 (30.52) | 278 (58.16) | 18 (3.67) | |
Past drinker | 67 (13.81) | 69 (14.44) | 72 (14.66) | |
Current drinker | 270 (55.67) | 131 (27.41) | 401 (81.67) | |
First, we retrieved the NNS data from the Food and Nutrition Research Institute (FNRI) of the Philippines based on mutual agreement, whereas the KNHANES data were freely accessible online via Korea Disease Control and Prevention Agency (KDCA) [
17]. Second, men, unmarried women (for the NNS only) and women with missing information on blood lipids and glucose were excluded from the NNS and KNHANES before the matching.
Demographic and anthropometric measures
Covariates associated with BP, lipids and glucose profiles included in this study were smoking, alcohol drinking, education level, occupation, BMI and waist circumference. Participants' smoking status was categorized into ever smoker and never smoker; ever smoker was defined as past or current smoker while never smoker was defined as never smoked at least 100 cigarettes in their entire lifetime. Alcohol drinkers were categorized into past or current drinker (including occasionally or had at least one to twelve alcoholic beverages of any type in their entire lifetime) or never drinkers (lifelong abstainers or who never consumed one or more drinks of any type of alcohol in their entire lifetime). Participants' employment status was categorized as employed and unemployed (including housewife, student, retired, etc.). In addition, participants were asked about their educational attainment where elementary or middle or high school level were grouped into high school and below, and those who had at least college education were grouped into college and above.
Collection of blood samples and determination of glucose and lipid profiles
For the FiLWHEL, the Cobas 8000 C702-1/C703-I analyzer (Roche Diagnostics, Basel, Switzerland) was used to determine serum levels of TC, TG, and HDL-C by enzymatic method and glucose by hexokinase UV assay. For the NNS, the TC, TG, HDL-C and glucose were assessed by enzymatic colorimetric method using the Roche COBAS Integra and Hitachi 912 analyzer. For the KNHANES, the TC and TG were analyzed by enzymatic method, HDL-C by homogeneous enzymatic colorimetric method, and glucose by hexokinase UV assay on Hitachi Automatic Analyzer 7600/7600-210 (Hitachi Co., Tokyo, Japan). All blood samples were collected by phlebotomists or trained medical personnel after at least 8–12-h overnight fasting through a venipuncture procedure.
The FiLWHEL coefficients of variation (CVs) were as follows: TC (1.43–2.24%), TG (1.85–2.10%), HDL-C (1.55–2.51%), and glucose (1.15–1.22%) [
16]. The blood samples for NNS participants were stored on ice and later centrifuged to separate plasma, which was later packed, labeled and frozen at − 20 °C until ready for analysis in the laboratory. The CVs for the NNS were directly derived and calculated from the 8th NNS clinical data and were as follows: TC (2.85–4.27%), HDL-C (1.44–5.08%), LDL-C (2.87–4.07%), TG (1.65–23.55%), and glucose (0.73–10.39%) [
18]. The KNHANES blood samples were processed and immediately refrigerated then transported in cold storage to the Central Testing Institute on the same day of the collection. The serum was separated immediately by centrifugation for analysis. The KNHANES CVs were as follows: TC (0.9–1.70%), HDL-C (0.9–2.06%), LDL-C (1.10–2.60%), TG (0.9–2.77%), and glucose (0.34–2.07%) [
20]. LDL-C was not directly assayed in the FiLWHEL study. The LDL-C was calculated from the measured values of the TC, TG and HDL-C using the Friedewald equation [LDL-C = (TC)-(HDL-C)-(TG)/5], where all values are expressed in mg/dL [
21].
Criteria for the definition of CVD risk factors
Obesity
BMI was calculated as weight (kg) divided by square of height (m
2), and respondents were classified according to the World Health Organization cut-off points: underweight, < 18.5 kg/m
2; normal, 18.5– < 23.0 (for Asian) and 23.0– < 25.0 kg/m
2; overweight, 25.0– < 27.0 kg/m
2; pre-obese, 27.0– < 30.0 kg/m
2 and obese, ≥ 30.0 kg/m
2 [
22]. Also, for waist circumference, we defined abdominal obesity using the absolute cut-off points from the 2001 NCEP (National Cholesterol Education Program) and 1998 NHLBI Obesity Education Initiative Expert Panel for Asian and American women (normal: 80– < 88 cm; abdominal obesity: ≥ 88 cm) classifying high waist circumference for each population in this study [
23].
Hypertension
Using the 2017 American College of Cardiology/American Heart Association (ACC/AHA) guidelines, systolic and diastolic blood pressures were classified as follows: normal, < 120/ < 80 mmHg; prehypertension, ≥ 120/ ≥ 80 mmHg or < 140/ < 90 mmHg; and hypertension, ≥ 120/ ≥ 90 mmHg [
24].
Dyslipidemia and elevated blood glucose
The 2005 American Heart Association/National Heart, Lung, and Blood Institute (AHA/NHLBI) criteria was used to categorize the blood lipids and glucose levels to determine dyslipidemia: high TC ≥ 200 mg/dL; low HDL-C < 50 mg/dL; high LDL-C ≥ 130 mg/dL; high TG ≥ 150 mg/dL, and elevated glucose ≥ 100 mg/dL [
25].
Statistical analysis
Continuous variables were presented as means ± SD and compared across the three populations (FiLWHEL, NNS, and KNHANES) using the analysis of variance (ANOVA), but categorical data were presented as n (%) and compared using Mantel-Haenzel test. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated using conditional multivariate logistic regression models with the women from the KNHANES as the reference population. Women with missing BMI, waist circumference, BP, lipid and glucose data were excluded when we analyzed these variables as outcomes in the logistic regression models. The multivariate analyses were adjusted for energy intake (kcal/day, continuous), smoking status (ever, never), alcohol intake (never, past, current), educational level (high school or below, college or above), employment status (employed, unemployed), BMI (< 18.5, 18.5– < 25.0, 25.0– < 30.0, ≥ 30.0 kg/m2) and waist circumference (< 85 cm, ≥ 85 cm) as applicable. Tukey's posthoc test was used for multiple comparison analyses to assess significant differences in nutrient intake across the population. All statistical analyses were performed using the SAS version 9.4 software package (SAS Institute Inc., Cary, NC, USA), and two-tailed P values < 0.05 were statistically significant.
Discussions
In this study, we compared CVD risk factors among FW in Korea and in the Philippines and KW. We found that FW in the Philippines had a higher prevalence of obesity and unfavorable lipid profiles than FW in Korea, suggesting that migration to Korea led to lower vulnerability to poor CVD health outcomes. However, FW in Korea had a higher prevalence of obesity and hypertension than KW.
The etiology of CVD burden among FW in the Philippines is complex and warrants further studies, but can be explained in the following ways. The average diet of a Filipino is characterized by less consumption of fruits and vegetables and higher consumption of meat, fast foods, fried foods and sugar-sweetened beverages [
26] along with the adoption of a Westernized diet [
27], responsible for the steady increase of obesity among Filipino in the Philippines for the past two decades from 20.2% in 1998 to 26.6% in 2019 [
18]. In this study, FW in the Philippines were found to have the highest prevalence of obesity across the population. Moreover, we found the highest intake of carbohydrates in FW in the Philippines among the three populations, which may be related to a low HDL-C. Genetic factors may be potential reasons for these differences. The Cebu Longitudinal Health and Nutrition Survey (CLHNS) showed genome-wide significant associations of several genetic variants with obesity [
28] and triglycerides and cholesterol [
29]. However, it remains unclear whether the genetic traits of Filipinos are key relevant factors for obesity, hypertension or dyslipidemia. Interestingly, when we compared FW in Korea with KW, we found a similarity in the lipid profiles, suggesting that changes in dietary habits and health practices [
30], an increase in food diversity [
31], and an improvement in socioeconomic status [
32] may be associated with a lower prevalence of dyslipidemia.
Several epidemiologic studies have reported the associations between dietary habits and lipid profiles [
33‐
35]. For example, a prudent diet consisting of a high intake of fruits, vegetables, seafood, whole cereal and low-fat dairy products was associated with reduced plasma TC (− 12.0 mg/dL), LDL-C (− 12.0 mg/dL), and apo B (− 6.6 mg/L) among women in Latin America [
33]. Also, adherence to a Mediterranean dietary pattern increased the HDL-C levels and decreased the TG/HDL-C ratio when compared to a Western dietary pattern [
34]. Similarly, healthy Korean diet (which is high in whole grains, legumes, nuts, vegetables, mushrooms, and fruits) improved the lipid profiles among Korean adults with type 2 diabetes [
35].
Immigrants in the US tend to adopt a more Westernized dietary pattern, a major contributor of growing rate of obesity, one of the prominent risk factors of CVD, in the US from 30.5% (1999–2000) to 41.9% (2017–2020) [
36]. In the current study, Filipino immigrant women in Korea who were exposed to a Korean diet and dietary practices had a lower cholesterol and TG level but had a higher mean HDL-C (58.19 ± 14.06 mg/dL) compared to the FW in the Philippines (HDL-C 40.26 ± 11.80 mg/dL). KW were exposed to a more varied diet compared to FW in the Philippines [
31,
37]. The Korean diet traditionally consists of cooked rice, dishes with broth, and small dishes (side dishes mostly seasoned or fermented), with a proportionally high vegetable intake, moderate to high legumes and fish intake, and low red meat intake [
38]. Individuals from different dietary traditions showed relatively stable dietary habits despite the occasional deviation, and accordingly, the prognosis of CVD risk factors might differ among communities, families, and socioeconomic strata [
11]. Studies have shown that exposure to a host culture may lead to changes in diet and health outcomes [
27]. However, the molecular and cellular mechanisms by which dietary components affect CVD risk factors may be complex and yet to be clearly understood [
39]. Further prospective studies are needed to evaluate the dietary mechanisms. We can also attribute this unexpectedly better lipid profiles of FW in Korea compared to FW in the Philippines to the healthy immigrant effect [
40]. In this current study, this theory might implies that FW in Korea are exposed to varied and healthier Korean diet regardless of their socio-economic status after they came. Although it is not difficult to adjust to different food cultures no matter how beneficial, FW in Korea might find it challenging. Although this healthy immigrant effect alone could not explain the reason for these disparities between immigrants and non-immigrants Filipino, it could at least broaden our understanding of the susceptibility and prognosis of CVD risk factors among FW in Korea. However, this theory need further study to warrant this account.
The reasons why we observed unfavorable lipid profiles among FW in the Philippines compared to Korean are not clear but warrants further studies. It is possible that the transition to unhealthy lifestyle factors may be associated with unfavorable lipid profiles among Filipino women [
11]. In the Philippines, the prevalence of elevated TC increased from 28.0% (2003) to 47.2% (2013). Similarly, the proportion of Filipinos with an elevated LDL-C profile increased by more than 15% between 2003 and 2013 from 31.5% (2003) to 47.5% (2013) and for low HDL-C from 54.2% (2003) to 71.0% (2013), particularly among women [
41].
We noticed that FW in Korea presented a higher prevalence of hypertension when compared to KW. A change in the food environment after migration might gradually change their dietary habit. Our study showed that FW in the Philippines had the lowest mean percent energy from fat, but after migrating, the mean percent energy from fat of FW in Korea was higher than KW. Koreans have one of the highest salt intake worldwide, almost half of which is derived from traditional diets such as salted vegetable (
kimchi) [
42]. Excessive sodium consumption significantly increased BP and CVD complication [
43]. Exposure to an unhealthy diet, including the Korean diet high in sodium and the traditional Filipino diet high in fat (e.g., deep-fried foods), could presumably be the culprit of the overall high prevalence of hypertension among FW in Korea. In addition to dietary factors, genetic factors may be related to an increase in BP. Several genetic variants predict the risk of developing hypertension among normotensive individuals in GenSalt study [
44]. Although there have been studies examining the genetic variants associated with hypertension among East Asians [
45], but such studies are scarce among Filipinos, and it is difficult to elucidate the complexity of genetics affecting hypertension among FW in Korea. Obesity is a risk factor for hypertension. However, adjustment for obesity appreciably attenuated the positive association among FW in the Philippines and in Korea, suggesting the mediation effect by obesity. The reason why we found a higher prevalence of obesity warrants further studies, including how socioeconomic segregation influences the susceptibility of women to obesity [
46].
Unhealthy dietary habits and the lifestyle of Filipinos may be important in the significant transitions of CVD events in the last decade. A recent review reported several risk factors of CVD prevalent among Filipino immigrants in the US. The study had indicated that Filipinos tend to develop an acculturated Westernized diet, which is associated with a higher risk for CVDs, hypertension, type 2 diabetes, and metabolic syndrome [
14]. The prevalence of CVD risk factors among Filipino immigrants in the US might due to higher rates of smoking (17.7%) [
12], lower levels of HDL-C (mean ± SD: 40.8 ± 0.2 mg/dL) [
47], overweight or obesity (BMI ≥ 25 kg/m
2, 60.4%) [
48], abdominal obesity (> 88 cm, 84.9%) and physical inactivity (48%) [
49], high carbohydrate, sodium and fat diet [
50,
51], and chronic conditions such as dyslipidemia (30.2%), diabetes (8.7%) and hypertension (41.2%) which are known CVD risk factors [
52]. Additionally, the prevalence of CVD risk factors among Filipino American women increased at a BMI as low as 23–24.9 kg/m
2 to ≥ 30 kg/m
2, including diabetes (HbA1c ≥ 6.5%, 6.38 ± 2.61 to 6.40 ± 1.08, respectively) and hypertension (SBP ≥ 140 mmHg, 125 ± 15 to 139 ± 19, respectively; or DBP ≥ 90 mmHg, 81 ± 8.4 to 87 ± 12, respectively), at a higher TG (≥ 150 mg/dL, 110 ± 55 to 142 ± 79, respectively), and at a lower HDL-C level (≤ 60 mg/dL, 60 ± 13 to 55 ± 14, respectively) [
53].
To the best of our knowledge, our study is the first to compare the susceptibility of CVD risk factors among Filipino and Korean women of diverse backgrounds matched by age (1:1 ratio). Studies conducted on Filipino immigrants in South Korea are sparse and out-of-date. Thus, the literature is limited to support the account regarding the Filipino immigrants’ health status. Another strength of our study is the multivariate adjustment for potential confounders. Some limitations of our study warrant consideration in the interpretation of the results. First, the cross-sectional design of our study delimits the inference of causality. Second, the FiLWHEL study has a relatively small sample size based on convenience sampling which may not represent the general population. Third, we did not adjust for physical activity even though it is a known risk factor of CVD because of the different questionnaires used in collecting physical activity information across the studies. However, adjustment of BMI or waist circumference may take into account the adjustment for physical activity levels to some degree.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.