The prevalence of anemia and iron deficiency among pregnant Ghanaian women, a longitudinal study

Background Gestational iron deficiency (ID) can be deleterious to mother and fetus. However, iron status is not routinely measured during pregnancy in Ghana. Therefore, the scope of ID in this population is unknown. Objective To determine the prevalence of anemia and ID across pregnancy in the Central Region of Ghana. Methods Women were recruited during their 1st trimester of pregnancy (< 13 weeks; n = 116) and followed through to their 2nd (n = 71) and 3rd (n = 71) trimesters. Data on socio-demographic variables, weekly intake of iron-rich foods and vitamin C-rich fruits were collected. Blood samples were drawn and the concentrations of hemoglobin (Hb), ferritin (Ft), serum iron (sFe), total iron binding capacity (TIBC), were measured; transferrin saturation (TSAT) was calculated. Repeated measures ANOVA was used to determine change in anemia and iron variables over time with groups categorized by 1st trimester iron status. Results Participants were 27.1 ± 5.2 years, on average. Prevalence of anemia (Hb <11.0 g/dL) was 37%, 63%, 58%; ID (Ft <15 μg/L) was 16%, 20%, 38%; and iron deficiency anemia (IDA; based on low Ft and Hb) was 6%, 12%, 25% in 1st, 2nd and 3rd trimesters, respectively. Significant changes in Hb, Ft and TIBC occurred across time. Iron status at 1st trimester had a significant effect on 2nd but not 3rd trimester iron status. Conclusions ID is prevalent in pregnant Ghanaian women, especially during the 3rd trimester. Anemia is a major public health problem during pregnancy in Ghana with a significant proportion due to factors other than ID.

previous antenatal attendances that were in-line with our research needs were purposefully selected. This yielded nine facilities. Simple random sampling was then used to select the health facilities used for the study. These facilities included the Moree Clinic, Cape Coast Teaching Hospital, the Cape Coast Metropolitan Hospital, Ewim Polyclinic, University of Cape Coast Hospital, Elmina Urban Health and Abura Dunkwa District Hospital. The nurses on duty informed prospective participants about the study and interested pregnant women were directed to the research team.

Screening and recruitment
A brief written screening form was used to determine eligibility, which included attendance at any of the seven selected prenatal clinics in Central Region of Ghana; aged between 18-38 years old at enrolment; <13 weeks gestation at enrolment (determined by last menstrual period or ultrasound scan); expecting a singleton pregnancy with no known congenital anomalies; and no known history of diabetes mellitus or hypertension. Eligible and interested participants completed the consent form (which was read to each participant, since some of the women had no formal education) and were recruited. Upon written informed consent (either providing a signature or a thumb print), socio-demographic characteristics, anthropometric and blood pressure assessments, and venous blood sample draw occurred immediately for the first trimester visit, unless the participant requested to come back at a later date. Four trained field assistants carried out data collection. After the first visit was completed, each participant was provided with a date for her 2 nd trimester visit. Participants were then followed into their 2 nd (13-27 weeks) and 3 rd (28-36 weeks) trimesters. At each health facility, a nurse was recruited to coordinate activities between patients and the trained data collectors. At the end of the first two visits, each woman received a bar of soap plus transportation cost as incentive, and at the end of the third visit, each woman received a baby onesie plus transportation cost.

Data collection procedure
Samsung Galaxy tablets were used to collect socio-demographic information including age, marital status, parity, education level, income and employment status. Dietary intake of vegetables, fruits and iron-rich foods, during the week prior to data collection, were also assessed at each visit.
Blood draws. Approximately 4 mL of blood were taken from the participant at each of the three trimesters during pregnancy by a trained phlebotomist at each health facility. Hb levels were determined on the spot via Hemocue (HB201; HemoCue America, Brea, CA, USA). The blood samples were stored on ice and delivered to the Cape Coast Teaching Hospital within 30 minutes of collection. Blood samples were centrifuged and serum aliquoted by a laboratory technician at the Cape Coast Teaching Hospital and subsequently stored in a -80 0 C freezer prior to shipping to The Pennsylvania State University, USA, where iron status biomarkers (serum iron (sFe), total iron binding capacity (TIBC) and serum ferritin (Ft)) and inflammatory markers (alpha-1-acid glycoprotein (AGP) and c-reactive protein (CRP)) were determined in Dr. Murray-Kolb's laboratory. Ft was determined via ELISA (Ramco Laboratories TX, USA), calibrated against WHO standards. sFe and TIBC were determined using colorimetric methods [13]. Transferrin saturation (TSAT) was calculated as (sFe/TIBC)×100. AGP and CRP were measured using radial immunodiffusion tests (Kent Laboratories Inc., Bellingham, WA, USA) and used to adjust Ft concentrations when inflammation was present. Ft values reported have therefore been adjusted for inflammation based on Thurnham criteria [14].
Follow-up visits for 2 nd and 3 rd trimesters. During the 2 nd (13-27 weeks) and 3 rd (28-36 weeks) trimester visits, all assessments, measurements and blood sample collection were repeated as described above, with the exception of socio-economic status (SES) assessment (which was only administered at the first visit).

Ethical approval
Ethical approval was obtained from the ethical review board of the Ghana Health Service Ethical Review Committee, University of Cape Coast Institutional Review Board, Cape Coast teaching Hospital Ethical Review Committee and The Pennsylvania State University Institutional Review Board.

Statistical methods
Statistical Analysis Software (SAS) version 9.4 (SAS Institute, Inc., Cary, NC, USA) was used for data analysis. Univariate analyses were run for all variables with the appropriate transformations applied to normalize all non-normal variables. The prevalence of ID was determined at each trimester. Proportions were presented for all sociodemographic (SES, marital status, age, years of schooling, income, parity) variables and iron supplement intake variables, and differences between the proportions of pregnant women who were iron deficient across the three trimesters were determined. One-way ANOVA and repeated measures ANOVA were used to determine significant differences between iron biomarkers over time and for change in iron variables over time based on 1 st trimester iron status, respectively. Simple regression models were used to determine if gestational age predicted iron status, adjusting for prenatal supplement intake.

Sample population
Two hundred and twelve pregnant women were screened, out of which 154 were eligible ( Fig  1). Thirty-five women refused participation so a total of 119 participants were recruited in their first trimester. Out of those recruited, 46 pregnant women dropped out of the study due to reasons such as miscarriage, spouse refusal, and unreachable (recipient out of coverage area and unanswered phone calls). Seventy-three pregnant women were followed into their 2 nd trimester. In the 3 rd trimester, 72 pregnant women were followed with 15 pregnant women dropping out between 2 nd and 3 rd trimesters due to reasons such as relocation, delivery and refusal to continue participation. Fourteen pregnant women from their 1 st trimester who were absent during the 2 nd trimester visit came back for the 3 rd trimester visit. The number of participants with complete socio-demographic and intake data were 116, 71 and 71, while 111, 68 and 65 participants provided a blood sample for 1 st , 2 nd and 3 rd trimesters, respectively. After removal of outliers and missing blood concentrations (not enough blood to measure every biomarker), 109, 65 and 60 participants had values for all iron biomarkers for 1 st , 2 nd and 3 rd trimesters, respectively (Fig 1).

Baseline characteristics of pregnant women
The baseline socio-demographic characteristics of the pregnant women are shown in Table 1. The majority of the women (56.9%) were recruited from the Komenda Edina Eguafo Abirem (KEEA) District, specifically from Elmina Urban Health Center, with the least recruited from the Abura Asebe Kwamankese District (12.9%). The average age of participants was 27 years with 26% having BMI in the overweight range and 11% classified as obese (Table 1). About 68% of participants were married with 80% living with their husbands/partners or co-existing. About 17% of these women were heads of their households, meaning they were the sole breadwinners of their household. About 8% of the women had no formal education, while 45% had up to a middle school education, with only 10% having a university degree. In terms of income, 79% of these women had an income-generating activity, and the majority of these women (84%) earned between 500-1000 Ghana cedis per month (equivalent to US $125-

Change in iron biomarkers over time
Significant changes occurred in Hb, Ft and TIBC across time (Table 2). No significant change was found for sFe or TSAT over time. The pattern of change varied depending on the iron biomarker in question. Women showed significantly higher Hb concentrations in their 1 st trimester (11.1 ± 0.1 g/dL) compared with 2 nd (10.6 ± 0.2 g/dL) and 3 rd (10.5 ± 0.2 g/dL) trimesters; 2 nd and 3 rd trimester Hb concentrations did not differ. Like Hb, Ft values decreased over time.
Pregnant women in their 1 st trimester showed significantly higher Ft concentrations (84.9 ± 5.6 μg/L) than pregnant women in their 2 nd (40.8 ± 7.2 μg/L) and 3 rd (27.6 ± 7.5 μg/L) trimesters, even after adjusting for inflammation. There was no significant difference in Ft concentrations between 2 nd and 3 rd trimesters. Consistent with iron status decreasing over time when measured by Ft, there was an increase in TIBC over time. Women in their 1 st trimester showed a significantly lower TIBC concentration (352.2 ± 8.1 μg/dL) than women in their 3 rd trimester (387.1 ± 10.7 μg/dL). Even though trends increased over time, there was no significant difference in TIBC concentration between 1 st and 2 nd trimesters nor between 2 nd and 3 rd trimesters. In a linear regression model with gestational age as a predictor of iron biomarker (Table 3), gestational age significantly predicted Ft and TIBC concentrations after adjusting for the number of prenatal supplements consumed. A unit increase in gestational age was associated with a decrease in Ft by 3.73 μg/L (p = <0.001) and an increase in TIBC by 1.96 μg/dL (p = 0.041). Gestational age was not a significant predictor of Hb, sFe or TSAT after adjusting for the number of prenatal supplements consumed.

Change in iron status over time depending on iron status at 1 st trimester
There was significant interaction between all iron biomarkers (Hb, Ft, sFe, TIBC, TSAT) and time categorized by 1 st trimester iron status (S1 Table,   at the 3 rd trimester, ferritin levels were comparable between pregnant women who were Ft sufficient and those who were deficient in trimester one (Fig 2). For sFe, pregnant women who were deficient in the 1 st trimester increased sFe concentrations significantly across trimesters while those who were sufficient in sFe at 1 st trimester did not experience a change in their levels across trimesters. At the 3 rd trimester, women who were sFe deficient and those who were sufficient in the 1 st

Discussion
In Ghana, not all biomarkers of iron status are routinely assessed during pregnancy and therefore the prevalence of iron deficiency in this population is unknown. We found that the prevalence of ID increased as pregnancy progressed with a high prevalence in the 3 rd trimester, as hypothesized. We found a significant time effect for Ft and TIBC concentrations, reflecting a pattern of ID which is commonly observed during pregnancy. As iron stores deplete, the amount of transferrin which is available to bind iron increases. Ft levels significantly decreased with gestational age while TIBC significantly increased with gestational age, as expected, during pregnancy. A limited number of studies have measured iron status during pregnancy in Ghana with one reporting a prevalence rate of iron deficiency ranging from 5-46%, depending on the definition applied [15], while another reported a rate of 11% for women �24 weeks and 20% for women �36 weeks pregnant, with an overall prevalence of 16% during pregnancy [16]. We observed prevalence rates of 16%, 20% and 38% in the 1 st , 2 nd and 3 rd trimesters, respectively. The major difference between these previously published studies and ours is that the previous studies were cross sectional while our study was longitudinal. Additionally, Mockenhaupt et al. [18] did not report prevalence rates by trimester and Engmann et al. [19] dichotomized gestational age into either �24 weeks or �36 weeks whereas we reported prevalence rates for 1 st (<13 weeks), 2 nd (13-27 weeks) and 3 rd (28-36 weeks) trimesters. Furthermore, Engmann et al. [19] used a population of women from the city and Mockenhaupt et al. [18] used a rural population; our population was predominately semi-urban. These differences may contribute to the various prevalence rates observed in our study compared to previous studies. A third study from Ghana which occurred in a comparable setting to the Central Region found the prevalence of ID to be 14%, 23% and 26% in the 1 st , 2 nd and 3 rd trimesters, respectively [17]. Several measures of iron status were used, as in our study; however, the authors did not indicate the definition used to determine ID. Not only were cut-offs not reported, but, whether ID was based on one or multiple iron biomarkers was not indicated. Their rates in the 1 st and 2 nd trimesters were comparable to our 1 st and 2 nd trimester rates (which were based on Ft<15 μg/ L) but our 3 rd trimester ID prevalence rate was higher (38%) than their 3 rd trimester prevalence of 26% [17]. As with the other two previous studies in Ghana, this study was cross sectional. Previous studies conducted in neighboring countries have reported higher ID rates during pregnancy. A study in Nigeria reported an ID prevalence rate of 48% [18], while a study in Malawi reported a prevalence rate of 32% [19] throughout pregnancy. Both studies were cross sectional, with the Nigerian study recruiting women at term while the Malawi study recruited only anemic pregnant women, thus excluding women who may have been iron deficient but not anemic.
When we compared the prevalence rates of ID found in our study to those in a high-income country such as the USA, we found a comparable pattern. Miller used the US NHANES 1999-2006 data and found an ID prevalence rate of 25.4% during pregnancy [20], while Mei et al. found ID prevalence of 7.3%, 23.7% and 29.2% for 1 st , 2 nd and 3 rd trimesters, respectively [21] using a Ft cut-off of <12 μg/L. If we use the same cutoff of 12 μg/L for our population, we see a higher prevalence in the 1 st trimester (12%), a slightly lower prevalence rate in the 2 nd trimester (18%), and a comparable prevalence rate in the 3 rd trimester (30%). This suggests that ID prevalence among pregnant women in Ghana is comparable to rates observed in the US but lower when compared with rates from neighboring low-and middle-income countries. One likely explanation is the use of iron supplements, especially during the 2 nd and 3 rd trimesters. This is the standard of care in Ghana where pregnant women are given 30-60 mg of iron daily from the 2 nd trimester until delivery. If a woman is found to be anemic before this time point, iron supplements are prescribed upon the diagnosis of anemia. In our population, 23% took iron supplements in the 1 st trimester, while 90% and 82% took iron supplements in the 2 nd and 3 rd trimesters, respectively. Another possible explanation for prevalence rates that are lower than neighboring countries could be the high consumption of meat and fish in this population [22]. Our population is located along the coast with fishing as the main source of employment and fish as the main source of protein in these communities [23]. In our study, we found that the frequency of consuming red meat increased throughout pregnancy and that most of the women regularly consumed fruits with a high vitamin C content. Consumption of meat and fish along with foods containing a high vitamin C content likely reflects a diet with higher iron bioavailability compared to the diets of neighboring countries.
Despite the fact that these women were given iron supplements as routine standard of care, Ft levels decreased across pregnancy. Other iron biomarkers such as TSAT and serum iron did not change significantly over time, perhaps as a result of the iron supplements given as standard of care. The rate of anemia was high among the women studied. A rate of 37% in the 1 st trimester is regarded as a public health problem. Higher rates were observed in the 2 nd (63%) and 3 rd (58%) trimesters, both classified as severe public health problems.
The prevalence of IDA was 6%, 12% and 25% for 1 st , 2 nd and 3 rd trimesters, respectively. Using the common assumption that half of anemia is due to iron deficiency [4], we would have expected rates of approximately 19%, 32%, and 29% for the 1 st , 2 nd , and 3 rd trimesters, respectively. Our lower than expected rates are an indication that much of the anemia in this population is due to factors other than iron deficiency. Although these rates are lower than those reported in several other countries [24], they are comparable to rates reported in other studies conducted in Ghana [19,20]. These findings challenge the WHO estimation that half of all anemic cases can be attributed to ID and remind us that careful examination regarding the etiology of the anemia is warranted, before ID is assumed and supplements are dispensed. This is especially important in low-and middle-income countries where the etiology of anemia is complex and the risk for poor outcomes of iron supplementation as a result of underlying infections could be significant. In this population, we cannot rule out other causes of anemia such as hemoglobinopathies, nutritional deficiencies (such as folate, vitamin B12 and vitamin A), infectious diseases (e.g. tuberculosis and HIV/AIDS), parasitic infestations (e.g. hook worm) and malaria [25]. As such, before one begins to treat anemia, it is important to consider the various causes and to determine an appropriate diagnosis before supplementation is started.
When we examined changes in iron biomarkers over pregnancy, based on iron status in the first trimester, we found a significant effect of iron status at first trimester on 2 nd trimester iron concentrations but not on 3 rd trimester concentrations. Generally, pregnant women who were iron sufficient at 1 st trimester had a drop in iron concentrations but still maintained a higher iron status in the 2 nd trimester than those who were iron deficient in the 1 st trimester. However, by their 3 rd trimester, both those who started pregnancy deficient and those who started pregnancy sufficient in iron had comparable iron status. This likely reflects the physiology of pregnancy in the 3 rd trimester where maternal iron is quickly mobilized in order to meet the fetal transfer demands, and the provision of iron supplements during the 1 st trimester to women diagnosed with anemia but not given to all others until the 2 nd trimester of pregnancy.
The strengths of our study include the longitudinal design with the assessment of iron status at three time points, using multiple iron biomarkers during pregnancy and the fact that the population studied included individuals whose iron status is not routinely assessed during pregnancy, therefore adding needed knowledge to the literature. The main study limitation was the high dropout rate (37.8%) observed between 1 st and 2 nd trimesters.

Conclusion
In Ghana, ID is prevalent during pregnancy, with the highest rates seen in the 3 rd trimester. Additionally, anemia remains a major public health problem during pregnancy in Ghana, and a significant proportion of anemia in this population is attributable to causes other than ID. Measures must therefore be put in place for thorough examination of anemia in pregnant women which should include assessment of iron biomarkers and not just Hb. This will help determine the cause of anemia before supplementation is started, which is especially important in countries like Ghana, where there are many potential causes of anemia.