Prevalence and Risk Factors for Iron Deficiency Anemia among Children under Five and Women of Reproductive Age in Pakistan: Findings from the National Nutrition Survey 2018

Introduction: Anemia remains a global public health problem, especially in developing countries. It affects primarily children under five (CU5), women of reproductive age (WRA), and pregnant women due to their higher need for iron. The most common form of anemia is iron-deficiency anemia (IDA). IDA is estimated to cause half of all anemia cases and one million deaths per year worldwide. However, there remains a lack of well-documented and biochemically assessed prevalence of IDA based on the representative population-based samples globally and regionally. In this study, we aimed to assess the National Nutrition Survey (NNS) 2018 to identify the prevalence and risk factors of IDA in Pakistani CU5 and WRA. Methods: Secondary analysis was conducted on the NNS 2018, a cross-sectional survey, which collected data on dietary practices, malnutrition, and food insecurity. Anemia was defined as hemoglobin levels < 11.0 g/dL in children and 12.0 g/dL in women. IDA was defined as low hemoglobin and low ferritin (<12 ng/mL) levels, adjusted for inflammation using AGP and CRP biomarkers in CU5 and WRA. Univariate and multivariable logistic regressions were conducted using Stata statistical software (version 16). We also compared the IDA rates of NNS 2018 and 2011. Results: A total of 17,814 CU5 and 22,114 WRA were included in the analysis. Of the CU5, 28.9% had IDA, while 18.4% of WRA reported to experience IDA. Among the CU5, IDA was most prevalent among male children aged 6–23 months living in rural areas and with the presence of diarrhea and fevers in the last 2 weeks. Children whose mothers had no education, were aged 20–34 years, and employed, had a higher prevalence of IDA. Married WRA, who are employed, living in rural areas, and with no education, had a higher prevalence of IDA. In the multivariable logistic regression, children aged 6–23 months (AOR = 1.19, 95% CI [1.08–1.33], p < 0.001) and with the presence of diarrhea in the last 2 weeks (AOR = 1.32, 95% CI [1.13–1.54], p < 0.001) or fever (AOR = 1.16, 95% CI [1.02–1.32], p = 0.02) had higher odds of IDA. At the household level, the odds of IDA among CU5 were higher in the poorest households (AOR = 1.27, 95% CI [1.08–1.50], p = 0.005), with ≥5 CU5 (AOR = 1.99, 95% CI [1.28–3.11], p = 0.002), and with no access to improved sanitation facilities (AOR = 1.17, 95% CI [1.02–1.34], p = 0.026). For WRA, the multivariable logistic regression found that the odds of IDA were higher among women with vitamin A deficiency (Severe: AOR = 1.26, 95% CI [1.05–1.52], p = 0.013; Mild: AOR = 1.36, 95% CI [1.23–1.51], p < 0.001), zinc deficiency (AOR = 1.42, 95% CI [1.28–1.57], p < 0.001), no education (AOR = 1.53, 95% CI [1.30–1.81], p < 0.001), and from severely food insecure households (AOR = 1.20, 95% CI [1.07–1.34], p = 0.001). The odds of IDA were lower among women whose body mass index was overweight (AOR = 0.77, 95% CI [0.69–0.86], p < 0.001) or obese (AOR = 0.71, 95% CI [0.62–0.81], p < 0.001). Conclusions: The child’s age, presence of diarrhea or fever, place of residence, household size, wealth status, and access to sanitation facilities were significantly associated with IDA among CU5 in Pakistan. For WRA, education, body mass index, vitamin A and zinc status, household food security status, wealth status, and access to sanitation facilities were significantly associated with IDA. Large, well-established, government-funded programmes focused on micronutrient supplementation, food fortification, the diversification of food supplies, and the treatment and prevention of infectious and parasitic diseases are needed to prevent IDA and all forms of anemia among children and women in Pakistan.


Materials and Methods
The National Nutrition Survey 2018, a cross-sectional survey, collected quantitative and qualitative data at the household level across the country [10]. A two-stage stratified sampling design was used, where the primary sampling units (PSUs) consisted of one village/city or part of a village/city containing an average of 200-250 households. These households were treated as secondary sampling units (SSU). The district-level representative sample size was calculated using the most recent prevalence of wasting, stunting, and micronutrient deficiencies in children, adolescents, and women. A total of 115,600 households were sampled to collect data on malnutrition, dietary practices, and food insecurity from all districts of Azad Jammu and Kashmir (AJK), Balochistan, Gilgit-Baltistan (GB), Islamabad Capital Territory (ICT), Khyber Pakhtunkhwa (including erstwhile Federally Administered Tribal Areas [FATA]), Punjab, and Sindh. Children under five (0-59 months), school-aged children (6-12 years), adolescents (10-19 years), and women of reproductive age (15-49 years) were the main target population of the survey. Additional details on the national survey design, methodology, and results are presented elsewhere [10].
Secondary data from 17,814 CU5 (3471 children aged 6-23 months and 14,343 children aged 24-59 months) and 22,114 WRA was used to identify the prevalence and risk factors of IDA by maternal age, education, and employment; child gender, age, and presence of disease; women's age, education, employment, and presence of disease; household socioeconomic status and place of residence. Univariate logistic regression was conducted to identify factors associated with IDA in CU5 and WRA. All data analyses were conducted in STATA version 16. Since IDA was the main outcome measure, women and children with results available for both hemoglobin and ferritin were included in the analysis. The frequencies, along with weighted percentages, were reported for selected predictors. The analysis started with simple univariate analysis followed by multivariate logistic regression. Unadjusted odds ratios with their 95% CIs were reported for the bivariate analysis. Variables significant at p < 0.25 were considered for inclusion in the multivariate model. Covariates that were not significant at the multivariate level were dropped from the model after careful assessment of confounding. The final model was selected based on theoretical and statistical significance of predictors. The Type 1 error was set at 0.05. The model estimates are presented with the adjusted odds ratios (AOR) and 95% CI.
The presence of anemia and IDA was assessed in CU5 and WRA via blood samples and an on-the-spot hemoglobin test using a HemoCue 301 machine. Anemia was defined as hemoglobin levels <11.0 g/dL in children and 12.0 g/dL in women [27]. IDA was defined as low hemoglobin and low ferritin (<12 ng/mL) levels, adjusted for inflammation using AGP and CRP biomarkers in CU5 and WRA. To evaluate CRP and AGP, micronutrient assays were conducted to measure plasma C-reactive protein and alpha-1-acid glycoprotein. Cobas c311, an automated chemistry analyzer (Roche, Basel, Switzerland), was used to quantify these proteins [28]. The analyzer performed an in vitro particle-enhanced immunoturbidimetric test that utilized latex to enhance the interaction between antigens and antibodies based on the immunological agglutination principle. The resulting complexes were then read turbidimetrically at specific wavelengths 660/340 nm for AGP and 546 nm for CRP. To ensure quality control, two levels of internal QC materials, Precicontrol Clinchem Multi levels 1 and 2 (Roche, Basel, Switzerland), were used at least once every twenty-four hours whenever a new assay kit or reagent was introduced or after performing a fresh calibration [29]. Any result falling outside the typical reference range was identified and rechecked immediately. The CRP values were reported to be the nearest tenth in mg/dL, with the AMR (Analytical Measurement Range) for the CRP assay being 0.1-0.25 mg/dL. The normal range for CRP was defined as 0-0.5 mg/dL. While for the AGP assay, the Roche/Hitachi Cobas c systems automatically calculated the concentration of the analyte in each sample and reported it in mg/dL. The AMR for the AGP assay was 10-400 mg/dL, where individuals with values greater than or equal to 100 mg/dL were considered to have elevated levels of AGP. To determine the plasma content of vitamin A (retinol), a reverse-phase high-pressure liquid chromatographic (HPLC) technique was employed. This analytical method helps separate and quantify different components in a mixture based on their chemical properties and retention times. The retinol levels were quantified using an Agilent 1200 series HPLC analyzer (Agilent, Santa Clara, CA, USA) with UV/Visible detection. To separate the components, a ZORBAX C-18 column measuring 4.6 × 150 mm with a particle size of 5 µm was utilized, and the analyte's concentration was measured at 325 nm. The mobile phase consisted of 100% methanol, maintained at a constant composition throughout the analysis. In order to correct for variable recovery, retinyl acetate was employed as an internal standard.
The measurement of plasma levels of 25-hydroxy vitamin D was performed using an automatic immunoanalyzer called Diasorin Liaison (Saluggia, VC, Italy). The technique employed for this analysis was a direct competitive chemiluminescence immunoassay (CLIA), which utilizes the principle of competition between the analyte (25-hydroxy vitamin D) and a labelled counterpart for binding to a specific antibody. Analytical Measuring Range (AMR) for the Liaison XL 25 OH Vitamin D assay is 4 to 130 ng/mL. Values < 20 were considered to be associated with vitamin D deficiency [30]. Plasma samples were analyzed for zinc estimation using an atomic absorption spectrophotometer, model ICE 3000 series (Thermofisher, Waltham, MA, USA). This technique relies on the fact that metals in their ground state absorb light at specific wavelengths. To achieve this, metal ions in a solution were converted to an atomic state through a flame. By supplying light at the appropriate wavelength (319 nm), the amount of absorbed light was measured against a standard curve. A flame with a flow rate of 0.9 to 1.2 L per minute (Air/acetylene) was used, providing a temperature of 2500 • C required for the atomization of gaseous zinc molecules within the flame. To ensure quality control, we employed serum controls from "UTAK Laboratories, Inc., Valencia, CA, USA" These controls were diluted with a specified volume of de-ionized water and remained stable for 30 days after dilution when stored between 2 and 8 • C. The controls were treated in the same manner as the analytical samples, meaning they were also diluted at a ratio of 1:10 with de-ionized water and then aspirated into the flame. A normal level of zinc in blood serum/plasma was considered to be less than 160 micrograms/dL [31]. A countrywide network of laboratories and collection centers was used to transport blood samples, maintain a cold chain, and ensure viability. The samples were analyzed at the Aga Khan University (AKU), Karachi, Sindh. The ethical clearance for the National Nutrition Survey and secondary analysis was obtained from Aga Khan University's Ethical Review Committee (ERC) and the National Bioethics Committee (NBC). All participants provided informed consent before survey enrolment. The secondary data analysis was approved by the ERC at AKU. Informed consent was obtained from all participants before the recruitment in this study, data collection, hematological investigations, and anthropometric measurements. The consent forms outlined the objectives, measurements, and process of confidentiality of this study.

Child, Maternal, and Household Characteristics
Out of the 17,814 CU5, 51.0% were male, with the majority aged 24-59 months (80.6%) and living in rural areas (62.1%) ( Table 1). The majority of the children CU5 were from Punjab (55.1%) and Sindh (27.2%), while 0.5% were from Gilgit Baltistan. A higher proportion of children had no reported presence of diarrhea (92.4%), acute respiratory infection (98.1%), and fever (86.3%) in the last 2 weeks. The mean Hb level of 10.6 ± 2.5 and ferritin level of 20.2 ± 32.3 was recorded among the CU5 enrolled in this study. Of the CU5, 28.9% were found experiencing IDA. More than 1 2 of the mothers of the children had no education (54.8%), while a 1 4 had completed secondary (12.4%) or higher education (10.4%). Most of these women were aged 20-34 years (71.5%) and were housewives (89.6%). Most households comprised ≤6 family members (57.7%), had <5 children under the age of five (99.1%), and had access to improved drinking water (92.4%) and improved sanitation facilities (84.4%). Furthermore, 60.8% of households reported being food secure, with 19.1% reporting severe food insecurity. A total of 5 out of 100 households (4.9%) reported receiving some form of financial assistance from the government in the last 12 months.

Women and Household Characteristics
Out of the 22,114 WRA, 93.1% were married, with the majority aged 30-49 years (55.1%) and living in rural areas (60.3%) ( Table 2). The majority of the WRA were from Punjab (53.7%) and Sindh (28.2%), while 0.5% were from Gilgit Baltistan. The mean Hb level of 12.1 ± 2.1 and ferritin level of 32.5 ± 61.1 was recorded among the WRA enrolled in the study. Of the WRA, 18.4% were found experiencing IDA. More WRA had a normal body mass index (45.4%; BMI 18.5-24.9 kg/m 2 ), normal vitamin A (72.9%), and zinc serum levels (77.8%). Additionally, about 79% of WRA were found to be deficient in vitamin D. More than 1 2 of the WRA had no education (53.7%), while a 1 4 had completed secondary (12.9%) or higher education (11.7%). Most of these women were housewives (86.5%), belonging to households with ≤6 family members (58.0%) and had access to improved drinking water (92.4%) and improved sanitation facilities (84.8%). Furthermore, 60.3% of households with the WRA were food secure, with 19.6% reporting severe food insecurity. A total of 5 out of 100 households (5.2%) reported receiving some form of financial assistance from the government in the last 12 months.

Risk Factors for Iron Deficiency Anemia in Children under Five
In the univariate logistic analysis, the odds of IDA were higher among children under two years of age (OR = 1.179, 95% CI [1.06-1.31], p = 0.002) with the presence of diarrhea (OR = 1.42, 95% CI [1.22-1.64], p < 0.001) or fever (OR = 1.26, 95% CI [1.12-1.42], p < 0.001) in the last 2 weeks compared with children aged 24-59 months with no diarrhea or fever (Table 3). Compared with children whose mothers had higher education, the odds of IDA were higher among children born to mothers with no education (OR = 1.203, 95% CI [1.03-1.40], p = 0.017) or primary levels of education (Primary: OR = 1.31, 95% CI [1.09-1.57], p = 0.005). Children with employed mothers (OR = 1.04, 95% CI [0.90-1.19], p = 0.617) had higher odds of IDA compared with children whose mothers were housewives but not significantly. In addition, the odds of IDA among children decreased with an increase in the mother's age; however, the odds ratios were not statistically significant for mothers aged 20

Risk Factors for Iron Deficiency Anemia in Women of Reproductive Age
In the univariate logistic analysis, the odds of IDA were higher among women from rural areas (OR = 1. 13

Comparison with Previous National Nutrition Survey 2011
We also compared the data for IDA in the NNS 2018 to the National Nutrition Survey conducted in 2011.
The data for children show that the overall IDA prevalence among children in Pakistan was reduced to 28.6% from 33.4%. Similar trends were seen in urban and rural figures. The maximum reduction was found in the Punjab province, whereas for the rest of the provinces, the IDA prevalence was increased (Figure 1).

Comparison with Previous National Nutrition Survey 2011,
We also compared the data for IDA in the NNS 2018 to the National Nutrition Survey conducted in 2011.
The data for children show that the overall IDA prevalence among children in Pakistan was reduced to 28.6% from 33.4%. Similar trends were seen in urban and rural figures. The maximum reduction was found in the Punjab province, whereas for the rest of the provinces, the IDA prevalence was increased (Figure 1). Similarly, when looking at the data for WRA, an overall reduction in IDA was observed from 19.9% to 18.2% in Pakistan. Similar trends were seen in urban and rural areas. Similarly, when looking at the data for WRA, an overall reduction in IDA was observed from 19.9% to 18.2% in Pakistan. Similar trends were seen in urban and rural areas. The The provincial data show a reduction in the prevalence of IDA in Punjab and Sind. However, in KP and Baluchistan, the prevalence had gone up (Figure 2).

Discussion
The prevalence of anemia among CU5 and WRA in Pakistan remains very high compared with global estimates [10]. Although, these prevalence are comparable with regional estimates and surpassed only by India (a country with a large vegetarian population) and Afghanistan (a country experiencing protracted conflict) [5][6][7][8][9][10][11][12][13][14][15][16][17][18]. The high prevalence of IDA among Pakistani children and women found in this study is in line with previous studies [32,33]. However, due to a lack of regional and global estimates determined using representative population-based surveys, the national prevalence of IDA cannot be reflected in detail [18].
This study presents selected risk factors associated with IDA among CU5 and WRA in Pakistan using the National Nutrition Survey 2018. This study shows children's age, the presence of diarrhea or fever, place of residence, household size, including the number of CU5, household wealth status, and access to sanitation facilities are significantly associated with IDA in CU5. Whilst for WRA, education, BMI, presence of vitamin A and zinc deficiencies, place of residence, household food security status, wealth status, and access to sanitation facilities were significantly associated with IDA. Risk factors associated with greater odds of IDA were children residing in Punjab and Sindh, with ≥5 CU5 experiencing diarrhea in the last 2 weeks and women with no education and deficient in vitamin A and zinc. Women and children from the poorest households, experiencing severe food insecurity and no access to sanitation facilities, had higher odds of IDA. The use of unsafe sanitation facilities can make these women and children more vulnerable to infectious and parasitic diseases and, hence, more likely to experience diarrhea, fever, and IDA [34]. Poverty, on the other hand, limits the ability of CU5 and WRA to access iron-rich nutritious foods. Household food security status was only associated with IDA in WRA; there was no association found between food insecurity and IDA in CU5. With seasonal food insecurity, a common occurrence across the country, women, mothers, and other female caregivers are accustomed to decreasing their portion sizes or skipping meals to ensure adequate food for other households' members, especially for children [35]. Gender inequality plays a significant role in contributing to anemia in Pakistan. Similar findings related to anemia have been reported within the region [33,[36][37][38][39]. In addition, the higher risk of

Discussion
The prevalence of anemia among CU5 and WRA in Pakistan remains very high compared with global estimates [10]. Although, these prevalence are comparable with regional estimates and surpassed only by India (a country with a large vegetarian population) and Afghanistan (a country experiencing protracted conflict) [5][6][7][8][9][10][11][12][13][14][15][16][17][18]. The high prevalence of IDA among Pakistani children and women found in this study is in line with previous studies [32,33]. However, due to a lack of regional and global estimates determined using representative population-based surveys, the national prevalence of IDA cannot be reflected in detail [18].
This study presents selected risk factors associated with IDA among CU5 and WRA in Pakistan using the National Nutrition Survey 2018. This study shows children's age, the presence of diarrhea or fever, place of residence, household size, including the number of CU5, household wealth status, and access to sanitation facilities are significantly associated with IDA in CU5. Whilst for WRA, education, BMI, presence of vitamin A and zinc deficiencies, place of residence, household food security status, wealth status, and access to sanitation facilities were significantly associated with IDA. Risk factors associated with greater odds of IDA were children residing in Punjab and Sindh, with ≥5 CU5 experiencing diarrhea in the last 2 weeks and women with no education and deficient in vitamin A and zinc. Women and children from the poorest households, experiencing severe food insecurity and no access to sanitation facilities, had higher odds of IDA. The use of unsafe sanitation facilities can make these women and children more vulnerable to infectious and parasitic diseases and, hence, more likely to experience diarrhea, fever, and IDA [34]. Poverty, on the other hand, limits the ability of CU5 and WRA to access iron-rich nutritious foods. Household food security status was only associated with IDA in WRA; there was no association found between food insecurity and IDA in CU5. With seasonal food insecurity, a common occurrence across the country, women, mothers, and other female caregivers are accustomed to decreasing their portion sizes or skipping meals to ensure adequate food for other households' members, especially for children [35]. Gender inequality plays a significant role in contributing to anemia in Pakistan. Similar findings related to anemia have been reported within the region [33,[36][37][38][39]. In addition, the higher risk of IDA among Punjabi and Sindhi CU5 and Sindhi and Pakhtun WRA may be linked with exposure to province-specific nutrition interventions and dietary habits.
Children aged 6-23 months were more at risk of IDA compared with older children (24-59 months) in this study. This finding is consistent with anemia-related studies conducted in Bangladesh, Bhutan, Nepal, and India [38,[40][41][42]. The higher risk of IDA among CU5 may be attributed to the inappropriate initiation of complementary feeding practices. This study also found that women with no education or lower levels of education had a higher risk of IDA compared to women with higher education. Similarly, several studies found that educated women were better informed and able to meet their iron requirements [36][37][38][39]43]. Furthermore, WRA with vitamin A and zinc deficiencies had higher odds of IDA, whereas women with overweight and obese body mass indices were less at risk of IDA compared to non-deficient and average body mass index WRA. Clinical studies have shown that anemia is influenced by vitamin A and zinc deficiency, which leads to impaired hemopoiesis, immunity, iron metabolism, and heme synthesis [44][45][46]. Yet, unlike this study, recent evidence shows that overweight and obese women experience chronic low-grade inflammation and high levels of serum hepcidin, which disturb iron absorption and increase the risk of anemia [47][48][49][50]. There is no clear justification for the lower risk of IDA among overweight and obese women shown in this study. A possible reason behind this protective effect may be higher socioeconomic standing and relative consumption of iron-rich sources.
Policies and programmes supporting micronutrient supplementation, food fortification, diversification of food supplies, and treatment and prevention of infectious and parasitic diseases (malaria, hookworms, etc.) are essential to address IDA (and all forms of anemia) among CU5 and WRA in Pakistan [21]. A multisectoral approach is needed, which also focuses on poverty reduction, female education, dietary diversity, and gender equality. Currently, the country is only providing iron-folic acid supplementation to pregnant women, and vitamin A supplementation and deworming to CU5 via the National Lady Health Worker Programme [22,23]. The programme has and continues to encounter significant financial and human resource issues, often relying on donor funds or in-kind donations of nutrition and health supplies to meet the population's needs. Supplementation with iron-folic acid for WRA and micronutrient powders for CU5 are usually provided via time-bound, donor-funded, and location-specific projects. The country has also implemented a large-scale food fortification programme (2016-2021) for wheat flour and edible oil with iron, folic acid, vitamin A, vitamin B12, and vitamin D [24][25][26]. This primarily donor-funded programme has built the capacity of millers via training and inkind provision of premix and micro feeders, strengthened regulatory and quality control infrastructures, and increased the demand and supply of fortified wheat and edible oil. Programme reach has been limited to a third of the country's population. Through strong political engagement in 2021, the programme has been able to legislate mandatory wheat flour and edible oil fortification in Sindh and Balochistan; while the remaining provinces rely on the 1965 national legislation on mandatory fortification of edible oil. With the programme concluding in 2021, the challenge of sustainability has come to the forefront, especially related to the in-kind provision of premix and micro feeders and the enforcement of legislation and standards. It is evident from our findings that these short-term and funding-restricted programmes and projects have been unable to meet the micronutrient needs of CU5 and WRA.
The major strength of this study is the use of consistent serum hemoglobin and serum ferritin, adjusted for inflammation using AGP and CRP biomarkers, to assess IDA among CU5 and WRA at the population level. Another strength is the use of the most recent, largest nationally representative sample, with a 90% response rate. Some limitations include the use of cross-sectional data, which cannot establish a causal relationship between IDA and the associated risk factors in CU5 and WRA. There have been suggestions that other biomarkers, such as hepcidin [51] and hemojuvelin [52], might improve the assessment of iron deficiency and responsive anemia, although this has not been borne out of trials using a targeted approach to supplementation with hepcidin [53]. Other studies have also suggested doing away with altitude adjustment [54] for anemia, but consensus in this aspect is lacking.

Conclusions
In conclusion, the study findings reflected that 28.9% of CU5 and 18.4% of WRA had experienced IDA. Regarding the risk factors of the IDA, this study found that the child's age, presence of diarrhea or fever, place of residence, household size, wealth status, and access to sanitation facilities are significantly associated with IDA among CU5 in Pakistan. For WRA, this study found education, body mass index, vitamin A and zinc status, household food security status, wealth status, and access to sanitation facilities are significantly associated with IDA. Large, well-established, government-funded programmes focused on micronutrient supplementation, food fortification, diversification of food supplies, and treatment and prevention of infectious and parasitic diseases are needed to prevent IDA and all forms of anemia among children and women in Pakistan.