Hibiscus sabdariffa meal improves iron status of childbearing age women and prevents stunting in their toddlers in Northern Ghana

Globally, iron deficiency (ID) is the most common form of nutritional deficiency, particularly in young children and childbearing age women. ID can lead to stunting and impaired cognitive development in children, as well as adverse maternal health and birth outcomes. In this study, the efficacy of an alternative food-to-food fortification utilizing indigenous iron-rich food sources was investigated in a quasi-experimental study. Childbearing age women (15–49 years, intervention-Kassena Nankana West district: n = 60; control-Builsa North district: n = 60) and their toddlers (6–24 months) consumed Hibiscus sabdariffa leaf meals (HSM, 1.71 mg Fe/100 g meal) three times a week for 12 weeks during the dry/lean season in Northern Ghana. We found that feeding the HSM (1.9 kg/day) improved iron status of women of childbearing age with time (p = 0.011), and protected stunting among toddlers during the dry/lean season (p = 0.024), which is the period with the worst food and nutrition insecurity. Compared with the control group, the number of stunted toddlers declined in the intervention group. This community-based 12-week feeding trial (clinical trials.gov ID: {“type”:”clinical-trial”,”attrs”:{“text”:”NCT03754998″,”term_id”:”NCT03754998″}}NCT03754998) was planned in a quasi-experimental design with the primary focus of assessing and improving iron status of dyads, determined by hemoglobin and sTfR. Baseline measurements (T1) were obtained prior to the intervention. Follow-up measurements (T2 and T3) were obtained six weeks and 12 weeks after T1, respectively. At the time of planning the current community intervention study (2015/2016), the PIs did not see the study fit under the classical definition of clinical trials that include any novel products (drugs, foods, instruments), but the definition has expanded since. Study site: This study was carried out in two districts in Upper East Region in Ghana (Figure 1): Kassena Nankana West (KNWD) and Builsa North (BND) districts. The capitals of these districts are about 25 km away from each other, are linked with a single paved road via the Kassena Nankana Municipality, and about 42 km away from the regional capital Bolgatanga. The study area has a typical savannah woodland vegetation characterized by short, scattered, drought-resistant trees and grass. The study area has two main seasons: Dry season (October–April/May), characterized by high temperatures, and rainy season (May/June–September). People in this area are mostly subsistence farmers. The study area is malaria endemic with peaks at the end of the rainy season when 45% of households are food insecure [20]. These districts were among the top five food insecure districts in the region. The selected sites for this study were two communities (Sakaa and Chania) in KNWD and three communities (Chuchuliga-yipaala, Azoayeri, and Awulansa) in BND. The study sites were selected based on the inclusion criteria of having a functional borehole (water source) throughout the dry season, existing women groups, and access to Community-Based Health Planning and Services (CHPS) compounds, sizeable number of mother and young children (6–23 months) dyads for good sampling frames, and community health nurses who were willing to work with researchers between May and August 2016. Researchers’ previous experience with those communities facilitated the community entry process. A map of the Upper East region showing its districts. The study subjects, women 15–49 years and their children 6–23 months, were recruited in May 2016. The dyads were drawn from selected districts using community-based birth registers at CHPS compounds kept by community health nurses stationed in these communities or by community health volunteers. The health volunteers are community members. Announcements through the community chiefs/leaders were then made in the respective communities for all women with children under five years of age to meet at their respective community health centers. The research team briefed the women and the community leaders on the study. Names that were shortlisted from the birth registers were read out. All dyads including those who were not in the register, but obtained from the health volunteers who knew almost everybody in their catchment areas, were contacted and checked on their willingness to participate in the study through the community health volunteers and community health nurses. Finally, verbal consent was sought from spouses of the women who were willing to be part of the study. A total of 120 dyads (60 mother/child dyads each for intervention and control group) were drawn from the two districts and agreed to participate in the study. Optimal design software (version 3.01) was used to calculate the minimal sample size of 100 dyads (50 dyads each for intervention and control groups) with a power of 80%, significance level of 5%, coefficient of determination of 65%, minimum detectable effect of 0.33 of diet on hemoglobin change, and 20% attrition rate. We had three sets of twins, one in intervention group and two in the control group in the study. We had a 7% attrition rate in intervention group. Two dyads relocated before baseline data collection and two relocated in the sixth week. All relocations were to either join a spouse or the entire nuclear family was migrating for farming purposes. Feeding trial: The participating dyads in intervention communities were invited to consume veo soup/meal (HSM) three times a week and were provided a weekly supply of iodized salt (450 g) for the household usage. Veo soup modified: The veo soup/meal is a local Ghanaian soup/meal mainly made of Hibiscus Sabdarifa leaves. It is a soup when it is prepared a bit watery and consumed with ‘tou zaafi’ (millet- or corn-based cooked paste). It is a meal when prepared thick and eaten by itself. The Hibiscus sabdariffa leaf meal (HSM) used in the present study was made of 18 kg Hibiscus sabdariffa leaves, 8 kg groundnut, 1.1 kg dawadawa (fermented African locust beans), 3 kg dried fish, plus 0.045 kg iodized salt, cooked with about 23 L (23 kg) water to yield 52.5 kg HSM. Dried fish or meat is a commonly added ingredient of HSM for people of high socioeconomic status, especially in the cities. The diet is named as HSM, as diets are usually labelled per their major ingredients. The ingredients of the HSM were provided to the women, who prepared the meals themselves. This helped in building trust and avoiding any forms of suspicion during supervision of all cooking and feeding activities. In each community, groups of ten women took turns to share the cooking activities, washing of bowls, and making water available for cooking. Three women in each community were trained on food weighing/measurement to conform to our standardized recipe. It took 45–60 min to cook a batch of the meal, which was then served (1.5 kg/woman) to all women and their children (0.5 kg/child) separately. The women and toddlers were given two separate bowls, so that the researcher could monitor each person’s consumption. All dyads were encouraged to consume to their satisfaction by requesting additional servings. The meal intake was measured by the researcher (CK) and trained women by the differences between the quantities served minus leftovers. The women and toddlers consumed, on average, 1.9 kg and 0.4 kg, respectively. The food composition analysis of the final HSM was carried out by Great Lakes Scientific, Inc. (Stevensville, MI, USA). Our standardized HSM (per 100 g) contained 1.7 mg iron, 6.6 mg protein, 4.6 mg fat, 82.6% water, and 2.7% ash. Questionnaire/interview: A questionnaire on food intake frequency and 24 h food intake recall was administered at baseline and at the end of the study. At baseline, information on sociodemographic characteristics was also taken. Malaria status was screened by the researcher at baseline, midpoint, and at the endpoint for all research participants using Rapid Malaria Diagnostic cassettes (Lot: 05CDB050DA. Standard Diagnostic, Inc. Republic of Korea.). Malaria screening is based on the presence of Histidine-Rich Protein-2 (HRP-II) in whole blood. HRP-II is known to be specific to Plasmodium falciparum, which causes more than 90% of malaria cases in Ghana [21]. The sensitivity and specificity of cassettes were 95% and 99.5%, respectively. All participants who were screened positive for malaria were referred to their respective CHPS compounds for treatment. In most instances, the community health nurses, who were in charge of compounds and also part of our research team, provided medicine to toddlers on site and women upon presenting their health insurance identification. Anemic individuals were also referred for treatment as ethically required, and potential interference or complication in intervention outcomes were handled during data analysis. Referred individuals were part of the study to the end, thus included in the data analyses. We modeled the Difference-in-Differences with continuous outcome measures (Hb and sTfR) and repeated measures ANOVA using the GLM procedure. This was done for ID and iron sufficient (IS) individuals at baseline and subsequently combined IS and ID. These analyses indicated non-significance. Interestingly, reporting rates of adequate and inadequate iron status is integral to our study analysis, so estimates of absolute risk differences are desirable. We used PROC GENMOD, a marginal population-level model (General Estimating Equations, or GEE) to generate these estimates in a binomial model using the identity link [22,23]. Anthropometry: Participants’ weights and heights were measured at baseline, midpoint, and at the endpoint, according to standard procedures [24]. Electronic scale (Serial number 5874030154862, Model 874 1321009. Seca gmbh & co kg, 22089 Hamburg, Germany.) was used to measure weights to the nearest 0.1 kg and height measured by Seca 217 (Seca gmbh & co kg, 22089 Hamburg, Germany.). Biochemical measurements: At baseline and endpoint, 5 mL whole blood was withdrawn into silica-coated serum separator vacutainers (Lot: 20140618, Anhui Kangning Industrial (Group) Co. Ltd, Tianchang City, China) and held at ambient temperature before and during transportation. Serum was separated using a centrifuge (Hettich) at 500× g for 5 min at room temperature. Separated serum was aliquoted, kept frozen at −18 °C (Thermo Fisher Scientific) at NHRC, then transported on dry ice to MDS Lancet laboratories in South Africa for serum transferrin receptor analysis. Analysis was done using Tina-quant Soluble Transferrin Receptor 80 tests, Roche/Hitachi cobas c 311, cobas c 501/502. Hemoglobin levels were measured in the field using HB 201 analyzer according to prescribed procedures. At midpoint, only hemoglobin levels were measured. Nutrient measurement: Iron concentrations of HSM, Hibiscus leaves, dawadawa, Amani, and groundnut were obtained by removing the organic content of samples in a high temperature muffle furnace. The resulting ashes were diluted in acid and absorbance was read by Atomic Absorption Spectrometry (3110 Perkin Elmer Atomic Absorption Spectrometer with a Hollow Cathode Calcium/Magnesium Lamp at 285.5 nm). Analyses were carried out by standard procedures for iron (AOAC 985.35), protein (AOAC 928.08II), fat (AOAC 925.12), moisture (AOAC 950.46A), and ash (AOAC 923.03). Analytical work was carried out by Great Lakes Scientific Inc. Stevensville, MI, USA. Proportion of iron from the various ingredients of HSM can be found in Table S1. Data analyses were carried out by SAS 9.4 (SAS institute, Cary, NC, USA). Characteristics of participants were described using frequency distributions. Comparison of participants’ characteristics between intervention and control groups was done using Chi-square statistics and Student t-test for categorical and continuous variables, respectively. Dependent variables of interest were linear growth (Stunting: HAZ < −0.2) of toddlers, and iron status (ID: sTfR > 4.40 and > 2.85 ug/L in women and toddlers Lab reference; anemia: Hb < 12 g/dL; IDA: concurrent presence of anemia and ID). Linear growth and iron status (measured by sTfR) were measured only at T1 and T3. T1, T2, and T3 are referred to herein as the time variable. Multiple logistic regression was fitted to identify the risk factors of iron status (ID), which is a binary variable. Since the response is discrete and correlated within the same subject, to account for dependencies, we modeled the relationship between the binary outcome and predictors using generalized estimating equations (GEEs) in PROC GENMOD. Control variables are household wealth index, marital status, sex of household head, mother's age, child's age, malaria status, participation in household decision-making, and the number of children and adults in household, which are measured at baseline study. Chi-square statistics was used to compare linear growth, IDA, and anemia status measured by hemoglobin levels in control and treatment groups. We compared the intra group hemoglobin levels using PROC MIXED, as we were interested in comparing or modeling means of hemoglobin levels. Statistical analysis for hemoglobin was done in twofold: 1) Analysis using nonanemic (Hb ≥ 12 g/dL) participants at baseline, and 2) analysis with anemic (Hb < 12 g/dL) plus nonanemic (Hb ≥ 12 g/dL) participants combined. It is worth noting that for iron status using sTfR, we had 58 samples at baseline in intervention, but one subject was excluded in our analysis due to inadequate sample volume for such analysis. At the endpoint, we had 53 results, which included endpoint results for the one missing at baseline, explaining the discrepancies in Table 1 and Table 2. The combined number of drop outs and missing data (12%) is less than the anticipated attrition rate of 20%. Based on the attrition rate of 12%, the final data analysis is not expected to be affected. Additionally, participants with missing variables were not significantly different from the other participants. Health indicators of childbearing age women (15–49 years) and their toddlers (6–23 months) in intervention and control groups across time. Iron deficiency (ID), defined by sTfR > 4.40 μg/L and >2.85 μg/L in women and toddlers, respectively. Anemia, defined by Hb < 12 g/dL and Hb < 11 g/dL, in women and toddlers respectively. Iron deficiency anemia (IDA), defined by concurrent presence of anemia and ID. Positive malaria: Have malaria. HAZ: Height-for-age z-score. WHZ: Weight-for-age z-score. Endpoint comparison of iron status among childbearing age women (15–49 years) in intervention and control groups. ID: sTfR > 4.4 μg/L. D-I-D: Difference in difference. sTfR-combined: Individuals with Hb ≥ 12.0 g/dL plus Hb < 12.0 g/dL. sTfR-combined: Individuals with sTfR ≤ 4.0 μg/L plus sTfR > 4.0 μg/L. Ethics: Research procedures were in accordance with the Michigan State University’s Institutional Review Board and Navrongo Health Research Centre (NHRC) Institutional Review Board (IRB) in Ghana