Source of data and samples

The data used in this analysis were derived from 2018 Basic Health Research data. Basic Health Research is a periodic community-based survey in Indonesia implemented since 2007 to collect basic data and health indicators regarding community health conditions at national, provincial, and district levels in Indonesia. Using a standardized questionnaire, the survey collects information on the community’s health status, along with its contributing factors, based on the model described by Henrik Blum [16].

The 2018 Basic Health Research consisted of 30,000 census blocks (CBs) distributed across all 34 provinces in Indonesia. Some of the health indicators collected in this survey represented the conditions at the district level. This survey employed a two-stage sampling design. In the first stage, CBs in urban and rural areas were used as the primary sample units. In the second stage, ten households from the target population of each CB were selected with equal probability using the systematic random sampling method. Household sampling was used for the interviews and blood sample collection. Of the 30,000 CBs, 2500 CBs were selected from 106 districts for blood tests to represent blood-related indicators at the national level [16]. The present study used information from 11,471 women respondents aged 19–49 years, who were not pregnant at the time of the interview and who resided in the households selected for the blood tests.

The validation of the questionnaire used in the 2018 Basic Health Research was carried out by the Indonesia Health Researchers Association. The pretesting of the questionnaire was conducted by researchers from the National Institute of Health Research and Development (NIHRD), Ministry of Health, prior to data collection.

The dependent variable was anemia in non-pregnant women of reproductive age (19–49 years). Anemia was defined as a hemoglobin level of <12 g/dL [28]. The survey used Hemocue Hb 201+ devices to measure the hemoglobin levels in the blood. The system consists of a battery-operated photometer and a disposable microcuvette coated with a dry reagent that serves as a blood collection device. For the test, a drop of capillary blood was taken from the tip of the respondent’s finger or heel into the microquin. Blood in the microcuvette was analyzed using the photometer [29].

The independent variables were BMI and MUAC. Bodyweight was measured using two types of digital scales. The first type has a capacity of 50 grams to 150,000 kilograms with an accuracy of 50 grams. The second type has a capacity of 5–150 kilograms with an accuracy of 100 grams. The weighing device was calibrated daily before data collection using four 1,5 liters of bottled water. The weighing scale was placed on a hard and flat floor. Respondents were required to remove any footwear, jackets, and the contents of any pockets before standing upright on the scales.

Body height was measured using a multifunctional height meter with a capacity of 2 meters and an accuracy of 0.1 cm. After setting up the height meter by mounting it on a flat and hard wall properly, the respondents stood with the feet flat on the floor, the heels against the corner where the wall and floor meet, with the head, shoulders, and buttocks touching the wall. Respondents were asked to stand up straight with eyes looking straight ahead, with the line of sight and chin parallel to the floor. Respondents must remove any footwear, headgear, hair ponytail, or bun. When the measurer was shorter than the respondents, the measurer could stand on a bench to ensure the measurement was conducted properly. BMI was divided into three categories (underweight: <18.5 kg/m2, normal weight: 18.5–24.9 kg/m2), and overweight and obese: >25 kg/m2) [30].

MUAC was measured using a measuring tape with an accuracy of 0.1 cm. The respondent should not hold anything with relaxed arm muscles with the left sleeve rolled up to reveal the shoulder base. The measurer then determined the midpoint between the respondent’s shoulder base and the elbow. The MUAC tape was wrapped around the arm at the midpoint mark, with the arm hanging straight down. MUAC was divided into two categories according to the UNICEF 2018 guidelines (low: <23.5 cm and normal: 23.5 cm) [16].

To maintain data quality, all enumerators held at least a diploma degree in a health-related field. Multiple training sessions were also carried out starting from the training of the masters of training, followed by the training of the trainers in Jakarta and eventually the training of the data enumerators in 34 provinces throughout Indonesia. All tools used in this survey were calibrated before taking measurements. Two enumerators carried out the measurement: the measurer and the recorder. All data were entered directly on-site using CsPro version 6.1. When the data management team identified outliers, enumerators were required to repeat the measurement. Furthermore, each district was supervised by a supervisor and an external validator team.

Other potential predictors

Our analyses included several potential predictors (women’s age, education level, physical activity, consumption of fruits and vegetables, and the presence of communicable or non-communicable diseases) that were considered confounding variables. The respondents were divided into three age categories (19–24, 25–35, and 36–49 years). The education variable was divided into three categories (primary school, junior high school, and university).

The respondents’ physical activity was measured out according to the modified Global Physical Activity Questionnaire (GPAC), which is part of the WHO STEPS instrument for measuring and monitoring risk factors for non-communicable diseases [31]. The metabolic equivalent of each task, which was the ratio of a person’s working metabolic rate relative to their resting metabolic rate, was used to determine if the respondent had sufficient physical activity per week [32]. Insufficient and sufficient physical activity was defined as metabolic equivalent of the task of <600 and ≥600 per week, respectively [33].

Fruit and vegetable consumption were measured based on the frequencies and serving sizes of fruits and vegetables ingested by household members aged ≥5 years. The instrument used to collect vegetable and fruit consumption data was based on the WHO STEPS [34]. In this analysis, we calculated the number of days that fruits and vegetables were consumed per week and the average number of servings per day. Sufficient and insufficient consumption were defined as the consumption of at least five or less than five servings of vegetables and fruits per day for seven days a week, respectively.

Regarding non-communicable diseases, the respondents were asked whether they had ever been diagnosed with asthma, cancer, diabetes mellitus, heart disease, hypertension, stroke, chronic renal failure, or joint disease by health professionals. Respondents who answered "yes" to any of these diseases were categorized as having a non-communicable disease. Similarly, respondents reporting having ever been diagnosed by health professionals with one of the communicable diseases were categorized as having a communicable disease. These communicable diseases were acute respiratory infections (in the last month), pneumonia (in the last year), pulmonary tuberculosis, hepatitis (in the last year), diarrhea (in the last month), malaria (in the last year), and filariasis (in the last 3 years).

Data analysis

In the first stage, a descriptive analysis was performed to examine the frequency distributions of all variables used in the analysis. Contingency tables were also used to examine the distributions of variables based on the women’s nutritional status. This analysis was followed by the bivariable and multivariable logistic regression analyses to assess the relationship between the combination of BMI and MUAC with anemia status, before and after adjusting for potential confounders. The multivariable logistic regression analysis included all variables that were significantly related to anemia status in the bivariable analysis.

Analyses were performed by considering the complex survey design, including the averaged weight variables, primary sampling unit, and strata. The amount of sample generated using complex sample analysis was defined as the number of weighted N. The final model used in this analysis was the model including all variables significantly associated with anemia status, using a significance level of 0.05 as the cut-off point. The odds ratios (ORs) and 95% confidence intervals (CIs) of all variables in the final model were reported. All analyses were performed using SPSS version 15 (SN: 6a45501054a0b0c57002).

Ethics approval

Ethics approval for the 2018 Basic Health Research of Indonesia was obtained from the Health Research Ethics Committee of the National Institute of Research and Development, Ministry of Health of the Republic of Indonesia (No: LB.02.01/2/KE.267/2017 and LB.02.01/2/KE.024/2018). All respondents were asked to sign an informed consent form before the interview was conducted and blood samples were collected.

Main findings

The results of our study showed that anemia remains a public health problem in Indonesia among non-pregnant women of reproductive age (19–49 years). We observed a significant association between anemia and women’s nutritional status, history of illness, and healthy behavior. Women with overweight and obesity were less likely to develop anemia compared to women with a normal BMI, regardless of their MUAC scores. We also found an increased likelihood of developing anemia in women with insufficient physical activity and sufficient fruit/vegetable consumption. Moreover, women diagnosed with a non-communicable disease had a reduced likelihood of developing anemia compared to women who had never been diagnosed with any illnesses. Our findings emphasize the need for effective health promotion interventions, including health education programs, to improve the nutritional status and healthy behaviors of women to reduce the prevalence of anemia and increase women’s health status in general.

Anemia and chronic energy deficiency in women of reproductive age

The high prevalence of anemia in non-pregnant women aged 19–49 years in this study indicated that anemia remains a public health problem [35]. Consequently, anemia is an important issue that requires serious attention, particularly as its prevalence has been relatively stagnant over time. In 2011, data from the World Health Organization (WHO) showed a prevalence of anemia in Indonesia of 22% (95% CI 12–37) [28].

Apart from anemia, women of childbearing age often suffer from chronic energy deficiency [36]. The 2103 Basic Health Research showed a prevalence of chronic energy deficiency in non-pregnant women of reproductive age of 14.5%, which increased to17.3% in pregnant women [17]. Studies from other countries have also reported a high prevalence of chronic energy deficiency in women of reproductive age [37, 38]. These findings signify the need for evidence-based and effective health interventions targeting all women of reproductive age to reduce and prevent anemia. Women, including adolescent girls, should be encouraged to consume iron supplementation, particularly during the menstrual period [39]. Moreover, health promotion programs to ensure the consumption of balanced and varied diets, including counseling by health workers and community workers, will also be beneficial [40, 41].

Factors associated with anemia in women of reproductive age

The association between anemia status and BMI and MUAC among the women included in our analysis is consistent with findings from other studies. A study conducted in 2013 reported a significant downward trend of anemia along with increased BMI; additionally, compared to women of normal weight, overweight and obese women also had lower odds of anemia [42]. Moreover, an inverse correlation between anemia and central obesity (PR = 0.75; 95% CI 0.63–0.89) was also reported. Similarly, a review on obesity and iron deficiency also reported low iron concentrations in the adult population with a high BMI [43]. Another study from Ethiopia also reported that MUAC of more than 23 cm reduced the incidence of anemia incidence by 0.41 [44].

A study conducted in Mexico, Peru, and Egypt reported similar findings [45]. The proportions of individuals with anemia in the group of women who were overweight (25 kg/m² ≤BMI <30 kg/m²) and obese (BMI> 30 kg/m²) were similar among the three countries. The prevalence of anemia in women from Egypt decreased as BMI increased. Although the difference between anemia prevalence was not significant across BMI groups among women in Peru, a decreasing trend of anemia in women was also observed with increasing BMI [45]. A high-calorie diet and adequate intake of animal-based food may increase hemoglobin formation in women with obesity [46]. A previous study with school children in China reported that although children with obesity had high hemoglobin levels, they could experience iron deficiency, which was not analyzed in our study [46].

Despite research results demonstrating a positive relationship between overweight and obesity with a reduced incidence of anemia, several studies have reported obesity as a cause of anemia in women [47]. Low serum iron levels are associated with weight gain and increased BMI [48]. Additionally, low iron levels and anemia also lead to fatigue and reduced physical activity contributes to weight gain [49].

Our findings showed a lower likelihood of developing anemia among women with non-communicable diseases compared to women who did not have any diseases. Previous studies reported increased risks of non-communicable diseases, such as diabetes mellitus, in individuals with excess iron in the body [50, 51]. We also postulated that the reduced risk of anemia in women with non-communicable diseases might be due to a high awareness of nutritional status and health conditions as a result of these women’s frequent encounters with health professionals. This frequent contact could help women to prevent nutritional-related problems and allow the early detection of health problems.

Furthermore, our results showed a stronger relationship of adequate physical activity and fruit and vegetable consumption with the occurrence of anemia. This finding is consistent with those of other studies reporting high risks of developing anemia among less active women [52]. Additionally, reverse “causation” might also contribute to this condition, as women with anemia might reduce their physical activities due to anemia signs and symptoms.

One explanation for the non-significant relationship between fruit and vegetable consumption and anemia was how this variable was constructed in our analysis. Adequate fruit and vegetable consumption was determined solely from the serving sizes of fruits and vegetables. The survey did not have information about the type of fruits and vegetables, which could be used to identify the consumption of iron-rich foods. In our study, the cut-off point used to define sufficient consumption was based on the WHO recommendations to prevent non-communicable diseases [53]. Therefore, our findings indicate the need to investigate the most appropriate cut-off point to prevent other health issues, such as anemia.

In general, our findings emphasize the necessity to strengthen iron supplementation programs for all women of reproductive age. Strategies to promote physical activity should be prioritized to improve community health behaviors. Our findings also demonstrate the need for a comprehensive assessment of the nutritional status of the women in the survey to help in the early detection of the double burden of malnutrition.

Strengths and limitations

This is the first study in Indonesia to investigate the relationship between combined BMI and MUAC and anemia in non-pregnant women of reproductive age. Since we used nationally representative data, our findings can be used to design and implement evidence-based interventions to prevent and reduce anemia at the national level. However, this study has several limitations. As in other cross-sectional surveys, the answers provided by the women were based on their recall ability. In the survey, data were collected manually and the calculation of anthropometric measurements such as BMI was conducted later after the fieldwork. Consequently, any inconsistencies between BMI and MUAC assessments could not be re-validated directly in the field at the time of data collection.