Dietary characteristics associated with hemoglobin and anemia among prospective first-time voluntary blood donors in Accra, Ghana Authors: Salmana Mehmood, Susan Telke, Edeghonghon Olayemi, Caitlin Ward, Lucy Asamoah-Akuoko, Bernard Appiah, Alfred Edwin Yawson, Tara Tancred, Amma Anima Benneh-Akwasi Kuma, Reena Ametorwo, Frank Ofori, Philip Baba Adongo, Jerry Daneku, Isaac Danso, Abigail Nana Afua Amegadoe, Catherine Segbefia, Cavan Reilly, Yvonne Dei-Adomakoh, Seth Adu-Afarwuah, the BLOODSAFE Ghana Investigators Published: 2026 ABSTRACT BackgroundOptimal dietary intake supports adequate hemoglobin (Hb) levels for blood donation. However, little is known about the dietary characteristics of prospective first-time voluntary blood donors (VBDs) in Ghana and their associations with Hb and anemia. We aimed to identify dietary characteristics associated with Hb concentration and anemia status among prospective first-time VBDs in Accra.MethodsThis cross-sectional study involved 223 participants who met all blood donation eligibility criteria, except the copper sulfate (CuSO4) screening test. Participants were over-sampled for low Hb. We collected background and dietary data, including number of meals consumed the previous day, foods usually consumed, and average frequency of consumption (AFC) of iron-rich foods and iron-absorption inhibitors (tea/coffee) using a questionnaire. Hb was measured using an automated Hematology Analyzer. We summarized dietary variables descriptively and analyzed associations using bivariate and multivariate statistics.ResultsMean age was 20 years. Most were females (63%) and students (98%). By participant report, 47% consumed ≤ 2 meals the previous day; 17% usually skipped breakfast; and 27% usually skipped lunch. Most reported high (weekly/daily) AFC of meat (78.9%), poultry (91.0%), fish (87.0%), fruits and vegetables (59.2%), legumes (72.6%), and tea/coffee (66.8%). Mean ± SD Hb was 12.4 ± 1.5 g/dL; anemia prevalence was 51.6% reflecting the deliberate oversampling for low Hb. In unadjusted models, reported animal source food (ASF) consumption at breakfast was associated with lower mean Hb (-0.57 g/dL; P = 0.016), while high AFC of poultry (+ 0.72 g/dL; P = 0.046) and legumes (+ 0.50 g/dL; P = 0.029) were associated with higher mean Hb; these associations were attenuated in multivariate models after adjusting for sex, age, and household assets index (HAI). In unadjusted analyses, high reported AFC of poultry was associated with lower odds of anemia (OR: 0.17; 95% CI: 0.05, 0.58; P = 0.005), and high AFC of tea or coffee with higher odds (OR: 1.95; 95% CI: 1.10, 3.44; P = 0.022); both associations remained significant after adjustment for sex, age, HAI, and ethnicity (poultry aOR: 0.16; 95% CI: 0.04, 0.66; tea/coffee aOR: 1.95; 95% CI: 1.05, 3.64).ConclusionHigh AFC of poultry was the most consistent dietary marker significantly associated with lower odds of anemia, whereas high AFC of tea or coffee was significantly associated with greater odds of anemia. Targeted nutrition interventions that address overall dietary diversity and iron bioavailability could help reduce anemia and improve donor eligibility in Ghana.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40795-026-01362-9. Introduction The availability of eligible blood donors is essential for blood transfusion in various medical situations, including emergencies, surgical procedures, and treatments [1]. However, in Africa, the rate of blood donation is often lower compared with the other regions of the world. For instance, of an estimated 118.5 million blood donations made worldwide in 2018, only 5% were from Africa, even though this region accounts for about 14% of the world’s population [2]. The World Health Organization (WHO) recommends repeat unpaid VBD over family or replacement donors [3]. The blood Hb level is a key factor in determining whether an individual is eligible to donate blood [4, 5]. Anemia or low Hb level leads to donor deferral, thereby reducing the overall availability of eligible blood donors and limiting the blood supply [6]. In 2021, the estimated prevalence of anemia across all ages and sexes in Ghana was 44.4%, which was considerably higher than the global prevalence of 24.3%. Much of this burden is often attributed to dietary iron deficiency, with vitamin A deficiency also contributing substantially [7]. Inadequate consumption of key nutrient-rich foods such as green leafy vegetables, legumes, meat, and fortified cereals [8], as well as nutritional enhancers such as vitamin C [9] and animal flesh [10] can lead to deficiencies in iron and other micronutrients [11] such as folate, vitamin A, and vitamin B12, resulting in impaired hemoglobin synthesis and anemia. Likewise, the presence of tannins and polyphenols in tea and coffee drinks can inhibit iron absorption from cereal-based iron-fortified foods [12] potentially contributing to anemia. We recently tested the non-inferiority of iron supplementation among first-time VBDs in Accra and found that the mean Hb concentration of participants with iron deficiency (ID) or anemia who received the intervention did not sufficiently match those in the control group (without ID or anemia and received no intervention) after 4 months to demonstrate non-inferiority, except when the analysis was restricted to females [13]. However, the dietary characteristics associated with Hb concentration and anemia status among the study participants remain to be investigated, despite the well-established relationship between dietary intake and anemia [8]. Knowledge of the dietary characteristics associated with Hb concentrations of these prospective first-time VBD could support the development of targeted nutritional interventions to improve the eligibility of these individuals for blood donation and enhance the overall blood supply in Ghana. The objective of the present study was to identify dietary characteristics related to Hb concentration and anemia status among prospective first-time VBDs who participated in a previous non-inferiority trial of iron supplementation in Accra, Ghana. Methods Study design, site, and participants This was a cross-sectional study involving 223 prospective first-time VBD who participated in a non-inferiority trial of iron supplementation and nutrition counselling under the BLOODSAFE Project, Ghana [13] These prospective first-time VBD were identified when they presented to donate blood at the static and outreach sites of the National Blood Service, Ghana (NBSG) Southern Zonal Blood Centre (SZBC), in Accra. As previously described [13], individuals (males and females) were eligible for the study if they met all NBSG blood donation criteria, regardless of their Hb screening results from the CuSO4 test. The NBSG criteria for blood donation were: between 18 and 60 years of age, weigh ≥ 50 kg, systolic and diastolic blood pressures between 90 and 140 mmHg and 60–90 mmHg, respectively; a pulse rate between 50 and 100 bpm; a non-contact forehead temperature not exceeding 37.5 °C; and meeting acceptable requirements for skin lesions, needle marks and physical appearance. Additional criteria were: negative test for malaria, no signs of helminthic infestation and, in the case of those who successfully donated, negative test for syphilis, HIV 1 & 2, hepatitis C, and hepatitis B. Potential participants who reported recently taking any iron supplements were excluded. Eligible potential participants, including those who did not pass the CuSO4 test (i.e., deferred from blood donation for low Hb), were recruited after research assistants explained the study procedures, answered any questions, and obtained informed written consent. To meet the objectives of the non-inferiority trial, individuals who did not pass the CuSO4 test were preferentially oversampled. Following consent, any participant found to have Hb < 10 g/dL at recruitment was withdrawn and referred for medical care. Data collection procedures Following consent, research assistants collected background information and administered a structured questionnaire. The questionnaire included several non-quantified food frequency items to collect data about dietary characteristics including (a) number of meals consumed the day before (ordinal response: 0 (fasting), 1, 2, 3 or more); (b) foods normally consumed for breakfast, lunch, and supper (open text response); and (c) average frequency of consumption (AFC) of selected iron-rich foods (meat, poultry, fish, fruits and vegetables, and beans and peas), in addition to tea and coffee drinks. The AFC of the selected iron-rich foods was assessed by using a five-point Likert scale (0 never, 1 rarely, 2 monthly, 3 weekly, 4 daily) for each. Approximately 5 ml of participants’ blood was collected by venipuncture into an EDTA tube for full blood count, including Hb assessment, using an automated Hematology Analyzer. Ethical considerations Ethics approval was obtained from the College of Health Sciences, Ethical and Protocol Review Committee, University of Ghana (protocol ID: CHS-Et/M.5-P4.14/2021). All study participants provided written informed consent. Description of dietary variables Variables describing the dietary characteristics of the participants were derived from the non-quantified food frequency questionnaire used to collect dietary information. As shown in S1 Table, we categorized the “Number of meals yesterday” as reported by participants as either low (< 3 meals) or high (≥ 3 meals). Participants’ responses for “Specify what you usually eat for: (a) breakfast, (b) lunch, (c) supper:” were classified into three main food groups for each meal: (i) animal-source foods (ASF), including meat, organ meat, poultry, eggs, milk and dairy products; (ii) iron-rich plant-source foods (PSF), including green leafy vegetables, other vegetables and fruits, and legumes; and (iii) iron absorption inhibitors (IAIs), mainly tea and coffee drinks. Participants were then categorized by whether they usually consumed ASF (yes/no), iron-rich PSF (yes/no), and IAI (yes/no) for breakfast, lunch, and supper. The AFC of each specific food item ((i) meat, (ii) poultry, (iii) fish, (iv) fruits and vegetables, (v) beans and peas, and (vi) tea and coffee), measured on a five-point Likert scale was dichotomized into low (0 never, 1 rarely, 2 monthly), and high (3 weekly, 4 daily). The AFC of tea and coffee (low vs. high) reflected the frequency of consumption of IAIs. Additionally, we summed the AFC responses for meat, poultry, and fish (MPF) to create an MPF score, with values ranging from 0 to 12, and AFC responses for fruits and vegetables, and beans and peas (FVB) to create an FVB score with values of 0 to 8. Subsequently, participants’ MPF score and FVB score were dichotomized based on the median values as low (< median) or high (≥ median). Importantly, the AFC variables and derived MPF and FVB scores represented self-reported frequency of consuming selected food groups and did not reflect quantified portion size, timing of consumption relative to meals, nutrient intake, iron density, bioavailability, or food preparation methods. Thus, these measures served as proxies for habitual exposure rather than direct indicators of nutrient intake or dietary adequacy. Statistical analysis We developed the statistical analysis plan before beginning the analysis, which was conducted using SAS Version 9.4 (SAS Institute, Cary, NC). Participants’ demographic characteristics, blood Hb concentration, and anemia (< 12.0 g/dL for females and Hb < 13.0 g/dL for males [14]) prevalence were summarized using means and standard deviations, or frequencies and percentages as appropriate. We calculated the percentages of participants by dietary variable categories. In bivariate and multivariate analyses of associations between dietary characteristics and Hb concentration or anemia status, we estimated unadjusted and adjusted mean (± SD or ± SE) Hb concentrations and anemia prevalence, by dietary variable categories. Covariates for the multivariate analyses were selected from eight background variables (age, sex, marital status, educational level completed, employment status, ethnicity, religion, and household assets index (HAI)), if they were associated with the outcome at p-value < 0.2 in bivariate tests of association [15]. The final multivariate model for hemoglobin concentration included sex, age, and HAI as covariates, while the model for anemia status included sex, age, HAI, and ethnicity. We adjusted only for these non-dietary covariates, and each dietary variable was evaluated in a separate regression model, which minimized the potential for multicollinearity among dietary indicators and avoided overadjustment. The composite MPF and FVB scores were also evaluated in models separate from those including their component food items. We did not conduct formal multicollinearity tests because overlapping predictors were not included simultaneously in models. The HAI was computed using Principal Component Analysis based on the ownership (yes/no) of 7 household items: television, radio, refrigerator, cookstove, car, bicycle, and motorbike. The first principal component was selected to represent the HAI, with higher values indicating greater asset ownership and, consequently, a higher socio-economic status (SES). All tests were two-sided, and statistical significance was determined at p < 0.05. Given that individuals who did not pass the CuSO₄ screening test were preferentially oversampled, the analytical sample included a higher proportion of individuals with low Hb than would be expected among first-time voluntary blood donors in Accra. Thus, the regression estimates described associations between dietary characteristics and Hb concentration or anemia status within this sample and were not intended to provide population-representative estimates of anemia prevalence. We did not apply sampling weights because our primary objective was etiologic, i.e., to examine whether dietary characteristics were associated with Hb and anemia among prospective donors. The oversampling of individuals with low Hb increased variability in the outcome and thus statistical power to detect these associations. Although this was a secondary analysis and not originally powered for the specific dietary associations of interest, the substantial number of anemia cases in our analytic sample supported multivariable logistic regression modeling. Furthermore, we calculated minimum detectable effect sizes at 80% power and a two-sided alpha of 0.05, rather than conducting post hoc power [16, 17]. With a fixed sample size of 223 and an assumed hemoglobin standard deviation of 1.5 g/dL, the study had 80% power to detect a minimum mean difference in hemoglobin ranging from 0.57 to 0.99 g/dL across the observed exposure group splits. For anemia status, assuming a reference prevalence of 58% [18], the study had 80% power to detect odds ratios of 0.47 or lower and 2.24 or higher for the most balanced dietary exposure split. Results The demographic characteristics of the study participants are shown in Table 1. The mean ± SD age was 20 ± 3 years. The majority (62.8%) were females, single (97.3%), and had no children (99.1%). Most (86.1%) lived with parents or relatives, used private transport (94.6%), and were students (97.8%). About one-half (50.2%) were of Akan ethnicity and the vast majority (95.1%) were Christians. A smaller percentage of participants owned a television (38%), a radio (37%), a refrigerator (36%), a cookstove (40%), or means of transport (9% for a car, motorbike, or bike). Table 1Demographic characteristics of study participantsBackground characteristicsn (%)Participants223Age, years; mean ± SD19.9 ± 2.8Sex Males83 (37.2) Females140 (62.8)Marital status Single217 (97.3) Married/Cohabiting6 (2.7)Had a child or children No221 (99.1) Yes2 (0.9)Current living arrangement Lived alone or in school26 (11.7) Lived with parents or relatives192 (86.1) Lived with partner only or partner and children5 (2.2)Primary means of transport Private211 (94.6) Public12 (5.4)Education level completed None1 (0.5) Junior high school135 (60.5) Senior high school78 (35.0) Diploma, Degree, Postgraduate9 (4.0)Current employment status Student218 (97.8) Informal/part-time / formal employment5 (2.2)Ethnicity Akan112 (50.2) Ewe44 (19.7) Dagbani/Hausa5 (2.2) Ga/Dangbe45 (20.2) Other17 (7.6)Religion Christian212 (95.1) Muslim11 (4.9)Ownership of household items Television set84 (37.7) Radio82 (36.8) Refrigerator81 (36.3) Cookstove89 (39.9) Car, motor-bike, or a bike19 (8.5) Table 1Demographic characteristics of study participantsBackground characteristicsn (%)Participants223Age, years; mean ± SD19.9 ± 2.8Sex Males83 (37.2) Females140 (62.8)Marital status Single217 (97.3) Married/Cohabiting6 (2.7)Had a child or children No221 (99.1) Yes2 (0.9)Current living arrangement Lived alone or in school26 (11.7) Lived with parents or relatives192 (86.1) Lived with partner only or partner and children5 (2.2)Primary means of transport Private211 (94.6) Public12 (5.4)Education level completed None1 (0.5) Junior high school135 (60.5) Senior high school78 (35.0) Diploma, Degree, Postgraduate9 (4.0)Current employment status Student218 (97.8) Informal/part-time / formal employment5 (2.2)Ethnicity Akan112 (50.2) Ewe44 (19.7) Dagbani/Hausa5 (2.2) Ga/Dangbe45 (20.2) Other17 (7.6)Religion Christian212 (95.1) Muslim11 (4.9)Ownership of household items Television set84 (37.7) Radio82 (36.8) Refrigerator81 (36.3) Cookstove89 (39.9) Car, motor-bike, or a bike19 (8.5) Demographic characteristics of study participants The percentages of participants who reported number of meals consumed during the day before the interview were 53% for ≥ 3 meals, 39% for two meals and 7% for one meal; only one participant (0.5%) reported consuming no meal. Figure 1 shows that for breakfast, the percentage of participants who reported usually consuming various food groups were 26% for ASF, 19% for iron-rich PSF (including green leafy vegetables, other vegetables and fruits, legumes), and 21% for tea or coffee, while about 17% reported usually consuming nothing or skipping breakfast. For lunch, 44% of participants reported usually consuming ASF, while the usual consumption of iron-rich PSF was reported by 49% of participants, with only a small percentage of participants (0.9%) reporting usually consuming tea or coffee, and 27% reporting usually skipping lunch. For supper, the percentage of participants who reported usually consuming ASF was 44%, while 61% reported usually consuming iron-rich PSF; only small percentages reported usually consuming tea or coffee (1.8%) or skipping supper (0.5%). Fig. 1Percentage of participants who reported usually consuming iron-rich animal source foods (ASF), iron-rich plant-source foods (Iron-rich PSF), tea or coffee (iron absorption inhibitors) or nothing for breakfast, lunch, and dinner Percentage of participants who reported usually consuming iron-rich animal source foods (ASF), iron-rich plant-source foods (Iron-rich PSF), tea or coffee (iron absorption inhibitors) or nothing for breakfast, lunch, and dinner The mean ± SD blood Hb concentration (g/dL) of the study participants based on the Hematology Analyzer results was 12.4 ± 1.5, with 51.6% of them being categorized as anemic (65.7% among females and 27.7% among males). Table 2 presents the percentages of participants by dietary variable categories, along with estimated unadjusted and adjusted mean (± SD or ± SE) Hb concentrations and anemia prevalence, by the dietary variable categories. Most participants reported a high (weekly or daily) AFC of meat (78.9%), poultry (91.0%), fish (87.0%), fruits and vegetables (59.2%) and legumes (72.6%). High AFC of tea or coffee was reported by 66.8% of participants. About one-half (51.6%) of the participants had a high MPF score, and likewise 51.6% had a high FVB score. Table 2Bivariate and multivariate analysis of the association between dietary characteristics and hemoglobin concentration and anemia status the study participants1CharacteristicsHemoglobin concentration, g/dlAnemia statusUnadjusted Mean ± SDUnadjusted Beta (95% CI)PAdjusted Mean ± SEAdjusted Beta (95% CI)2PUn-adjusted %Unadjusted OR(95% CI)PAdjusted %Adjusted OR (95% CI)3PUsually consumed ASF for breakfast 0 = no (74.0%)12.5 ± 1.512.7 ± 0.147.940.7 1 = yes (26.0%)12.0 ± 1.5-0.57(-1.02, -0.11)0.01612.4 ± 0.1-0.33(-0.66, 0.00)0.05362.11.78(0.96, 3.30)0.0752.11.58(0.81, 3.10)0.18Usually skipped breakfast 0 = no (83.4%)12.4 ± 1.612.7 ± 0.150.043.4 1 = yes (16.6%)12.2 ± 1.4-0.27(-0.82, 0.27)0.3212.5 ± 0.2-0.14(-0.54, 0.26)0.4959.51.47(0.71, 3.01)0.3052.91.46(0.66, 3.23)0.34Usually consumed PSF for supper 0 = no (39.0%)12.3 ± 1.512.6 ± 0.156.348.4 1 = yes (61.0%)12.4 ± 1.60.17(-0.24, 0.59)0.4112.7 ± 0.10.01(-0.29, 0.32)0.9348.50.73(0.42, 1.26)0.2642.30.78(0.43, 1.42)0.42AFC of meat 4 0 = Low (21.1%)12.3 ± 1.712.6 ± 0.255.348.5 1 = High (78.9%)12.4 ± 1.50.04(-0.46, 0.54)0.8712.7 ± 0.10.02(-0.34, 0.38)0.9150.60.83(0.43, 1.58)0.5643.50.81(0.40, 1.65)0.56AFC of poultry 4 0 = Low (9.0%)11.7 ± 1.312.4 ± 0.385.076.6 1 = High (91.0%)12.4 ± 1.50.72(0.01, 1.42)0.04612.7 ± 0.10.32(-0.20, 0.83)0.2348.30.17(0.05, 0.58)0.00534.90.16(0.04, 0.66)0.011AFC of fish 4 0 = Low (13.0%)12.3 ± 1.512.6 ± 0.262.155.6 1 = High (87.0%)12.4 ± 1.50.15(-0.46, 0.75)0.6312.7 ± 0.10.02(-0.42, 0.46)0.9250.00.61(0.27, 1.37)0.2343.40.61(0.25, 1.51)0.29AFC of fruits/vegetables 4 0 = Low (40.8%)12.3 ± 1.512.5 ± 0.153.949.9 1 = High (59.2%)12.4 ± 1.60.09(-0.32, 0.51)0.6612.8 ± 0.10.26(-0.04, 0.57)0.0950.00.86(0.50, 1.47)0.5741.40.71(0.38, 1.31)0.27AFC of legumes4 0 = Low (27.4%)12.0 ± 1.412.6 ± 0.157.446.2 1 = High (72.6%)12.5 ± 1.60.50(0.05, 0.96)0.02912.7 ± 0.10.09(-0.25, 0.42)0.6149.40.73(0.40, 1.32)0.2944.10.92(0.48, 1.78)0.80AFC of tea/coffee4 0 = Low (33.2%)12.6 ± 1.512.7 ± 0.140.533.5 1 = High (66.8%)12.3 ± 1.6-0.29(-0.72, 0.14)0.1912.6 ± 0.1-0.12(-0.44, 0.19)0.4457.11.95(1.10, 3.44)0.02249.61.95(1.05, 3.64)0.036MPF score5 0 = Low (48.4%)12.2 ± 1.412.6 ± 0.156.547.9 1 = High (51.6%)12.5 ± 1.70.34(-0.06, 0.75)0.1012.7 ± 0.10.10(-0.20, 0.40)0.5147.00.68(0.40, 1.16)0.1641.20.76(0.43, 1.36)0.36FVB score5 0 = Low (48.4%)12.2 ± 1.512.5 ± 0.156.549.0 1 = High (51.6%)12.6 ± 1.60.40(0.00, 0.81)0.05012.8 ± 0.10.21(-0.08, 0.51)0.1647.00.68(0.40, 1.16)0.1641.80.75(0.42, 1.34)0.331Total n = 223. AFC average frequency of consumption, ASF animal-source foods, FVB fruits, vegetables, beans, and peas, HAI household assets index, MPF meat, poultry, and fish, PSF plant-source foods2Adjusted for pre-specified covariates (sex, age, HAI) independently associated with hemoglobin concentration at P < 0.2 [15] in bivariate analysis. HAI was computed using principal component analysis based on the ownership (yes/no) of a television, radio, refrigerator, cookstove, car, bicycle, and motorbike; the first principal component was selected to represent the HAI, with higher values indicating greater asset ownership and, consequently, a higher socio-economic status3Adjusted for pre-specified covariates (sex, age, HAI, and ethnicity)) independently associated with anemia status at P < 0.2 in bivariate analysis4Low: never, rarely, monthly; High: weekly or daily5Low; < sample median value; High; ≥ median sample value Table 2Bivariate and multivariate analysis of the association between dietary characteristics and hemoglobin concentration and anemia status the study participants1CharacteristicsHemoglobin concentration, g/dlAnemia statusUnadjusted Mean ± SDUnadjusted Beta (95% CI)PAdjusted Mean ± SEAdjusted Beta (95% CI)2PUn-adjusted %Unadjusted OR(95% CI)PAdjusted %Adjusted OR (95% CI)3PUsually consumed ASF for breakfast 0 = no (74.0%)12.5 ± 1.512.7 ± 0.147.940.7 1 = yes (26.0%)12.0 ± 1.5-0.57(-1.02, -0.11)0.01612.4 ± 0.1-0.33(-0.66, 0.00)0.05362.11.78(0.96, 3.30)0.0752.11.58(0.81, 3.10)0.18Usually skipped breakfast 0 = no (83.4%)12.4 ± 1.612.7 ± 0.150.043.4 1 = yes (16.6%)12.2 ± 1.4-0.27(-0.82, 0.27)0.3212.5 ± 0.2-0.14(-0.54, 0.26)0.4959.51.47(0.71, 3.01)0.3052.91.46(0.66, 3.23)0.34Usually consumed PSF for supper 0 = no (39.0%)12.3 ± 1.512.6 ± 0.156.348.4 1 = yes (61.0%)12.4 ± 1.60.17(-0.24, 0.59)0.4112.7 ± 0.10.01(-0.29, 0.32)0.9348.50.73(0.42, 1.26)0.2642.30.78(0.43, 1.42)0.42AFC of meat 4 0 = Low (21.1%)12.3 ± 1.712.6 ± 0.255.348.5 1 = High (78.9%)12.4 ± 1.50.04(-0.46, 0.54)0.8712.7 ± 0.10.02(-0.34, 0.38)0.9150.60.83(0.43, 1.58)0.5643.50.81(0.40, 1.65)0.56AFC of poultry 4 0 = Low (9.0%)11.7 ± 1.312.4 ± 0.385.076.6 1 = High (91.0%)12.4 ± 1.50.72(0.01, 1.42)0.04612.7 ± 0.10.32(-0.20, 0.83)0.2348.30.17(0.05, 0.58)0.00534.90.16(0.04, 0.66)0.011AFC of fish 4 0 = Low (13.0%)12.3 ± 1.512.6 ± 0.262.155.6 1 = High (87.0%)12.4 ± 1.50.15(-0.46, 0.75)0.6312.7 ± 0.10.02(-0.42, 0.46)0.9250.00.61(0.27, 1.37)0.2343.40.61(0.25, 1.51)0.29AFC of fruits/vegetables 4 0 = Low (40.8%)12.3 ± 1.512.5 ± 0.153.949.9 1 = High (59.2%)12.4 ± 1.60.09(-0.32, 0.51)0.6612.8 ± 0.10.26(-0.04, 0.57)0.0950.00.86(0.50, 1.47)0.5741.40.71(0.38, 1.31)0.27AFC of legumes4 0 = Low (27.4%)12.0 ± 1.412.6 ± 0.157.446.2 1 = High (72.6%)12.5 ± 1.60.50(0.05, 0.96)0.02912.7 ± 0.10.09(-0.25, 0.42)0.6149.40.73(0.40, 1.32)0.2944.10.92(0.48, 1.78)0.80AFC of tea/coffee4 0 = Low (33.2%)12.6 ± 1.512.7 ± 0.140.533.5 1 = High (66.8%)12.3 ± 1.6-0.29(-0.72, 0.14)0.1912.6 ± 0.1-0.12(-0.44, 0.19)0.4457.11.95(1.10, 3.44)0.02249.61.95(1.05, 3.64)0.036MPF score5 0 = Low (48.4%)12.2 ± 1.412.6 ± 0.156.547.9 1 = High (51.6%)12.5 ± 1.70.34(-0.06, 0.75)0.1012.7 ± 0.10.10(-0.20, 0.40)0.5147.00.68(0.40, 1.16)0.1641.20.76(0.43, 1.36)0.36FVB score5 0 = Low (48.4%)12.2 ± 1.512.5 ± 0.156.549.0 1 = High (51.6%)12.6 ± 1.60.40(0.00, 0.81)0.05012.8 ± 0.10.21(-0.08, 0.51)0.1647.00.68(0.40, 1.16)0.1641.80.75(0.42, 1.34)0.331Total n = 223. AFC average frequency of consumption, ASF animal-source foods, FVB fruits, vegetables, beans, and peas, HAI household assets index, MPF meat, poultry, and fish, PSF plant-source foods2Adjusted for pre-specified covariates (sex, age, HAI) independently associated with hemoglobin concentration at P < 0.2 [15] in bivariate analysis. HAI was computed using principal component analysis based on the ownership (yes/no) of a television, radio, refrigerator, cookstove, car, bicycle, and motorbike; the first principal component was selected to represent the HAI, with higher values indicating greater asset ownership and, consequently, a higher socio-economic status3Adjusted for pre-specified covariates (sex, age, HAI, and ethnicity)) independently associated with anemia status at P < 0.2 in bivariate analysis4Low: never, rarely, monthly; High: weekly or daily5Low; < sample median value; High; ≥ median sample value Bivariate and multivariate analysis of the association between dietary characteristics and hemoglobin concentration and anemia status the study participants1 -0.57 (-1.02, -0.11) -0.33 (-0.66, 0.00) 1.78 (0.96, 3.30) 1.58 (0.81, 3.10) -0.27 (-0.82, 0.27) -0.14 (-0.54, 0.26) 1.47 (0.71, 3.01) 1.46 (0.66, 3.23) 0.17 (-0.24, 0.59) 0.01 (-0.29, 0.32) 0.73 (0.42, 1.26) 0.78 (0.43, 1.42) 0.04 (-0.46, 0.54) 0.02 (-0.34, 0.38) 0.83 (0.43, 1.58) 0.81 (0.40, 1.65) 0.72 (0.01, 1.42) 0.32 (-0.20, 0.83) 0.17 (0.05, 0.58) 0.16 (0.04, 0.66) 0.15 (-0.46, 0.75) 0.02 (-0.42, 0.46) 0.61 (0.27, 1.37) 0.61 (0.25, 1.51) 0.09 (-0.32, 0.51) 0.26 (-0.04, 0.57) 0.86 (0.50, 1.47) 0.71 (0.38, 1.31) 0.50 (0.05, 0.96) 0.09 (-0.25, 0.42) 0.73 (0.40, 1.32) 0.92 (0.48, 1.78) -0.29 (-0.72, 0.14) -0.12 (-0.44, 0.19) 1.95 (1.10, 3.44) 1.95 (1.05, 3.64) 0.34 (-0.06, 0.75) 0.10 (-0.20, 0.40) 0.68 (0.40, 1.16) 0.76 (0.43, 1.36) 0.40 (0.00, 0.81) 0.21 (-0.08, 0.51) 0.68 (0.40, 1.16) 0.75 (0.42, 1.34) 1Total n = 223. AFC average frequency of consumption, ASF animal-source foods, FVB fruits, vegetables, beans, and peas, HAI household assets index, MPF meat, poultry, and fish, PSF plant-source foods 2Adjusted for pre-specified covariates (sex, age, HAI) independently associated with hemoglobin concentration at P < 0.2 [15] in bivariate analysis. HAI was computed using principal component analysis based on the ownership (yes/no) of a television, radio, refrigerator, cookstove, car, bicycle, and motorbike; the first principal component was selected to represent the HAI, with higher values indicating greater asset ownership and, consequently, a higher socio-economic status 3Adjusted for pre-specified covariates (sex, age, HAI, and ethnicity)) independently associated with anemia status at P < 0.2 in bivariate analysis 4Low: never, rarely, monthly; High: weekly or daily 5Low; < sample median value; High; ≥ median sample value In the bivariate analyses, participants who reported usually consuming ASF for breakfast had significantly lower mean Hb concentration (unadjusted β = -0.57 g/dL; P = 0.016) than those who did not. Participants with high AFC of poultry (unadjusted β = +0.72 g/dL; P = 0.046) or legumes (unadjusted β = +0.50 g/dL; P = 0.029) had significantly higher mean Hb concentrations compared to those with low AFC. A high FVB score (≥ median sample value) was associated with borderline higher mean Hb concentration (unadjusted β = +0.40 g/dL; P = 0.050). In multivariate models adjusting for sex, age, and HAI, these associations were attenuated: the inverse association with reported ASF consumption at breakfast became marginally non-significant (adjusted β = -0.33; P = 0.053), while the associations with high AFC of poultry and legumes, as well as high FVB score, were no longer statistically significant. No other dietary characteristics were significantly associated with mean Hb concentration. In the unadjusted analyses involving anemia, participants with high AFC of poultry had 83% lower odds of anemia (OR: 0.17; 95% CI: 0.05, 0.58; P = 0.005), while participants with high AFC of tea or coffee had 95% higher odds of anemia (OR: 1.95; 95% CI: 1.10, 3.44; p = 0.022). These associations remained significant after adjustment for sex, age, HAI, and ethnicity. All other associations with anemia were not statistically significant in either the unadjusted or adjusted analyses. Discussion This study examined dietary characteristics associated with Hb concentration and anemia status among prospective first-time VBDs in Accra, Ghana. Participants reported frequent consumption of animal- and plant-source foods at main meals, though meal skipping and consuming two or fewer meals daily were common. Anemia prevalence was high (51.6%) reflecting the deliberate oversampling for low Hb. Among the dietary characteristics examined, high AFC of poultry was the most consistent marker, showing significant associations with higher Hb concentration and lower anemia risk. In contrast, high AFC of tea or coffee was associated with increased odds of anemia. This study had several strengths, including thorough dietary data collection, objective measurement of blood Hb concentration using an automated Hematology Analyzer, which enhanced the reliability of the findings, and recruitment of both sexes, which increased the representativeness of the sample by sex. A possible weakness was the reliance on participants’ recall for dietary data collection without the use of quantified dietary components, which may have introduced bias and limited precision in the dietary assessment. Furthermore, involving prospective first-time VBDs who met all blood donation eligibility criteria, except the CuSO4 test for Hb screening limits the generalizability of the findings to the broader population of first-time VBDs in Ghana, although the study was not intended to generate population-level estimates of anemia prevalence. Given that the sample was originally recruited for a separate non-inferiority trial with a different primary objective, no specific power calculation was conducted a priori to detect associations between dietary characteristics and Hb concentration or anemia. However, because most first-time VBDs tend to meet the pre-blood donation eligibility criteria aside from the CuSO4 Hb test, this potential weakness may have had only a limited impact on the generalizability of our findings. The background characteristics of the participants in our sample might provide important context for explaining the observed dietary characteristics, mean Hb concentration, and anemia prevalence. Given that the participants were predominantly young, female, and single, with most having no children, living with parents or relatives, and being students, this demographic may be more likely to engage in irregular meal patterns, such as skipping breakfast, due to factors such as convenience, financial constraints, and limited cooking skills [19]. With nearly all the participants (97.8%) being students, factors such as academic pressures, busy schedules, and peer influence could have contributed to meal skipping, particularly at breakfast and lunch [20]. Furthermore, the fact that most lived with parents or relatives suggested a sheltered lifestyle with limited autonomy over food choices. An unexpected finding was that usual ASF consumption at breakfast was associated with lower mean Hb (significant in the unadjusted model and marginal after adjustment). This inverse association may reflect residual confounding (e.g., unmeasured socioeconomic factors or broader dietary patterns correlated with breakfast ASF) rather than a causal effect of ASF itself. Reverse causality may also be possible, if individuals with perceived low Hb or fatigue modified their breakfast choices. In addition, breakfast ASF in this setting may include foods with relatively low iron content or bioavailability and likely small portion sizes, which we did not quantify; this may help explain why the association was observed for mean Hb but not for anemia status. The observed anemia prevalence in this sample should be interpreted with caution, as the study was not designed to produce a population-representative estimate. Individuals who did not “pass” the CuSO₄ test were intentionally oversampled to meet the objectives of the primary trial. As a result, the sample included a higher proportion of participants who were likely to have low Hb concentration, which could bias the anemia estimate upward relative to the general population of prospective first-time VBDs in Accra. However, the oversampling of individuals with low Hb supported our primary analytical aim of etiologic association estimation by increasing variability in Hb and anemia status. Therefore, while the regression analyses are informative for identifying associations between dietary characteristics and Hb or anemia within this group, this sample is not suitable for estimating population-level anemia prevalence among all first-time VBDs in Accra. Nonetheless, the anemia prevalence is comparable to that (58%) observed in our previous study of a purposive sample of 325 first-time VBDs from southern Ghana [18], for which model-based estimates adjusting for selection bias indicated that 61.6% of females and 19.7% of males were anemic by WHO criteria. The anemia rate in the sample was substantially lower than the 81% prevalence reported from an earlier study of VBDs deferred by the CuSO4 test in Accra [4]. The higher prevalence reported in the earlier study [4] may be due to differences in the study population, including greater mean age (30 years) of the sample and the inclusion of repeat donors, as well as the fact that all participants were deferred from blood donation due to low Hb from the CuSO4 test. A study by Cable et al. [21] showed that repeat donations could potentially lead to iron deficiency or suboptimal Hb production due to restricted iron availability. A few other key findings from our study warrant discussion. First, the observation that high AFC of poultry was significantly associated with higher mean Hb concentration (unadjusted) and lower odds of anemia (both models) suggests the relevance of frequency of consumption of this food as a dietary marker for Hb levels and anemia status. Poultry is a food source of iron [22], as well as vitamin B-12 [23] and high-quality protein [24], all of which are important for erythropoiesis. However, our AFC variables capture reported frequency of consumption rather than portion size, iron density, or detailed preparation methods. Furthermore, poultry consumption may reflect broader contextual factors (e.g., affordability, accessibility, or overall meal regularity) rather than iron intake alone. The lack of significant association of AFC of meat or fish with Hb concentration or anemia in our study may reflect smaller portion sizes or lower actual intake per serving compared with poultry [25], which were not captured by our non-quantified frequency measures. Second, the observed significant association between high AFC of tea or coffee (reflecting frequent consumption of IAIs) and increased odds of anemia in unadjusted and adjusted models is consistent with the biological plausibility that tannins and polyphenols in these beverages can inhibit non-heme iron absorption [26, 27]. It may be that participants often consumed tea or coffee with meals or within 1–2 h afterward, when such inhibitory effects are expected to be strongest [28–30]. However, our data on tea or coffee were limited to frequency of consumption and did not capture portion size, timing relative to meals, or preparation methods, so we cannot draw mechanistic conclusions about these pathways in this population. Nutritional interventions would be needed to address poor dietary practices and thereby increase mean blood Hb concentration or reduce anemia prevalence among prospective first-time VBDs. While promoting the frequent consumption of poultry could be part of an important dietary strategy in this population, such consumption alone is unlikely to be sufficient to prevent anemia without broader dietary improvements. Modifying dietary habits, such as consuming tea or coffee between meals [31] or including iron absorption enhancers such as vitamin C-rich foods in the diet [26, 32] can counteract the inhibitory effects of tannins in tea and coffee [33]. Vitamin C is known to act by reducing ferric iron (Fe3+) to ferrous iron (Fe2+), thereby enhancing the bioavailability of non-heme iron [32]. Finally, iron supplementation programs combined with education on improved dietary diversity [34] can be targeted to potential VBDs to help reduce deferral rates due to low hemoglobin. Conclusion Among the prospective first-time VBDs in Accra over-sampled for low Hb, high AFC of poultry was the most consistent dietary marker, significantly associated with greater mean Hb concentration and lower odds of anemia, whereas high AFC of tea or coffee was associated with higher odds of anemia. Poultry consumption may reflect broader contextual factors rather than just iron intake; targeted nutrition interventions that address overall dietary diversity and iron bioavailability could help reduce anemia and improve donor eligibility in Ghana. Supplementary Information Supplementary Material 1: Table S1. Description of derivation of dietary variables (DOCX). Supplementary Material 1: Table S1. Description of derivation of dietary variables (DOCX).