Large –scale wheat flour folic acid fortification program increases plasma folate levels among women of reproductive age in urban Tanzania

There is widespread vitamin and mineral deficiency problem in Tanzania with known deficiencies of at least vitamin A, iron, folate and zinc, resulting in lasting negative consequences especially on maternal health, cognitive development and thus the nation’s economic potential. Folate deficiency is associated with significant adverse health effects among women of reproductive age, including a higher risk of neural tube defects. Several countries, including Tanzania, have implemented mandatory fortification of wheat and maize flour but evidence on the effectiveness of these programs in developing countries remains limited. We evaluated the effectiveness of Tanzania’s food fortification program by examining folate levels for women of reproductive age, 18–49 years. A prospective cohort study with 600 non-pregnant women enrolled concurrent with the initiation of food fortification and followed up for 1 year thereafter. Blood samples, dietary intake and fortified foods consumption data were collected at baseline, and at 6 and 12 months. Plasma folate levels were determined using a competitive assay with folate binding protein. Using univariate and multivariate linear regression, we compared the change in plasma folate levels at six and twelve months of the program from baseline. We also assessed the relative risk of folate deficiency during follow-up using log-binomial regression. The mean (±SE) pre–fortification plasma folate level for the women was 5.44-ng/ml (±2.30) at baseline. These levels improved significantly at six months [difference: 4.57ng/ml (±2.89)] and 12 months [difference: 4.27ng/ml (±4.18)]. Based on plasma folate cut-off level of 4 ng/ml, the prevalence of folate deficiency was 26.9% at baseline, and 5% at twelve months. One ng/ml increase in plasma folate from baseline was associated with a 25% decreased risk of folate deficiency at 12 months [(RR = 0.75; 95% CI = 0.67–0.85, P<0.001]. In a setting where folate deficiency is high, food fortification program with folic acid resulted in significant improvements in folate status among women of reproductive age. The Tanzania National Food Fortification Program mandates fortification[28] to targeted staple foods including salt, edible oil, wheat and maize flour Table 1. With the exception of salt, which has been fortified with iodine since 1994, fortification of other food staples officially started in 2013 through a mandate passed by the government of Tanzania requiring all industrial processed wheat, maize and edible oils to be fortified. This study is limited to wheat flour fortification only since folic acid fortification of maize flour had not been implemented to scale at the time of the study. aWS = Water-Soluble Helen Keller International worked in partnership with the Ministry of Health and Social Welfare through Tanzania Food and Drug Authority (TFDA), the Tanzania Food and Nutrition Centre (TFNC), the Ministry of Industry and Trade, the Tanzania Bureau of Standards (TBS), and food producers to assist in the rollout of this mandate. To date 14 large-scale food producers participate in the national program, including all ten of the country’s wheat flour producers, and four large-scale vegetable oil refineries. Production capacity for fortified products among these 14 industries amounts to approximately one million metric tons (MT) of wheat flour and 300,000MT of vegetable oil annually, or roughly 88% and 80% of the market share, respectively. TFDA works to ensure fortified products meet the levels of quality required by standards set by Tanzania Bureau of Standards (TBS) at production point, in the market, and at ports of entry through regular inspections. Using the program impact pathway adopted from a similar evaluation undertaken in Costa Rica S1 File.[29], we mapped an impact pathway for the Tanzania Food Fortification program and identified the scope and key areas of focus for this evaluation (Fig 1). Adopted with modifications from Reynaldo Martorell et al. Am J Clin Nutr 2015; 101:210–217. a,bAreas of focus for this evaluation in the context of Tanzania National Food Fortification program. We conducted a prospective cohort study comparing participant’s plasma folate levels before and after the rollout and scale up of the national food fortification program. We enrolled non-pregnant women of reproductive age (18–49 years) living in the Temeke and Ilala districts of Dar es Salaam. These two districts were selected to provide a mix of urban and peri-urban populations assuming these communities would benefit from relatively faster access to fortified foods, given their close proximity to the main food industries in Dar es Salaam. To maintain maximum representation with high internal and external validity, participants were recruited from 10 clinics, 5 from each of the two selected districts, providing a 90% aggregate coverage of antenatal care for pregnant women within the population catchment area. Given the high rates of antenatal clinic (ANC) coverage by these clinics, we assumed that eligible participants attending Mother and Child Health (MCH) clinics from these facilities are a representative sample for women of reproductive age in the two districts selected for this study. One district hospital and four health centers MCH clinics were selected from each, Temeke and Ilala districts. We invited women attending clinics with their children for immunization program appointments to participate in this study. We randomly selected women to participate through a lottery system using identical folded cards with “YES” and “NO” labels. Women who picked “YES” were further screened for the study. An equal number of participants were enrolled across the ten participating sites giving a total of 60 participants from each site. Since no pre-existing data on prevalence of folate deficiency and effect size from the region could be found for reference, our sample size was calculated using estimates obtained from similarly designed studies in Latin America[30]. Assuming a two-sided alpha of 0.05, a power of 0.8, intraclass correlation of 0.02, and mean plasma folate levels of 10±7 and 13±9 ng/ml at baseline and end point, respectively, we obtained a sample size of 416 participants. In order to account for loss to follow up in this study, we inflated the sample size by roughly 50%, resulting in a total sample of 600 women of reproductive age. We enrolled 600 out of 827 women screened across the 10 study sites in Dar es Salaam (Fig 1). Inclusion criteria was based on participants not being pregnant, based on the date of their last menstrual period as well as a urinary test during screening, having given birth at least 9 months prior to study inclusion, planning to remain in the study area for the next 12 months, and giving written informed consent to participate in the study. We excluded participants who were currently or had previously taken folic acid/iron supplements during the past 9 months, and anyone with a reported chronic illness (Fig 2). The plasma folate status of enrolled participants was assessed at baseline, and at 6 and 12 months. During these visits, fasting blood samples of ten hours or greater were obtained to determine the plasma folate levels. To maintain contact with our participants and to enhance follow-ups, we conducted home visits and phone calls to all participants during months where no clinic visits were scheduled to provide nutrition counselling on nutrient rich diverse diets, use of fortified foods water and sanitation practices. We also collected dietary intake data from those who have missed their clinic visits as well as reminding participants on their upcoming clinic visit. Written, informed consent for voluntary participation was obtained from each study participant before enrolment. The consent process was conducted in Kiswahili, the local language. The research protocol and the consent forms were reviewed and approved by the Medical Research Review Council of the National Institute for Medical Research (MRRC–NIMR), Tanzania and the Institutional Review Board of the Harvard T. H. Chan School of Public Health, USA. In accordance to Tanzania standard of care, nutritional counselling was provided to all participants. A dietary intake assessment using a semi-quantitative FFQ, validated for this setting was conducted at baseline, and at 6 months, and 12 months of follow up to allow estimation of total intake as well as isocaloric comparison and interpretation of the plasma folate results [31]. The questionnaire was comprised of 108 food items commonly consumed. Participants were asked if they had consumed these foods in the prior month, and if so, how often, and the frequencies were converted to servings per day. A serving is based on food specific portion sizes, using the Tanzania Food Composition Tables[32]. Using this table, we computed the food folate, which is comparable to the dietary folate equivalents (DFE)[33] as well as total energy consumed. We restricted data to FFQs with reasonable total energy intake, >600kcals and <4500kcals. Venous blood specimens (about 4 ml) were collected in purple top tubes containing K2EDTA using standard venepuncture procedures, and samples were sent to the Africa Academy of Public Health (AAPH) laboratory within two hours of collection. Once in the laboratory, specimens were centrifuged at 3200 rpm for ten minutes to obtain plasma that was stored in a -20C freezer in 1 to 3ml vials. Laboratory scientists made sure that specimens were free from hemolysis, lipemia and icterus. Plasma folate assays were done using the Cobas e411 automated analyzer (Roche Diagnostics, Switzerland) as per manufacturer instructions. The machine was calibrated by lyophilized human plasma with folate. Adding 1.0 ml of distilled water carefully dissolved the contents of the calibrators. The calibrators were mixed carefully, avoiding foam formation. Aliquots of the reconstituted calibrators were transferred into empty vials and stored at -20°C till needed. Whenever required, a pair of calibrators were brought out and thawed before use at room temperature. Testing proceeded after successful calibration, with 250–300μl of each sample run in duplicate and average readings recorded. Negative and positive controls were used for quality assurance in each run. Controls were run individually at least once per 24 hours while the test was in use, once per new reagent kit and after each passed calibration. Printed results were certified by one laboratory personnel and reviewed by another one, before entered into a database. The normal range for the equipment is 1.5–20.0 ng/ml. We analysed change in mean plasma folate levels from baseline, to 6, and 12 months of follow-up. Mean (±SD) values of total energy intake (in calories/day), macronutrient proportions of the total calorie intake, as well as intake of fortified wheat-based foods (in servings/day) were calculated to estimate folic acid intake at baseline. We estimated the change in levels of these measures from baseline, and assessed statistical significance using a paired Student’s T-test. The concentrations suggested for defining folate deficiency based metabolic indicators range from 3–4 ng/ml [15,16,34]. This value is derived from data related to preventing anaemia and hyperhomocysteneinemia and the public significance of applying same cut-off in isolating folate deficiency in the context of NTDs is not fully understood[15,34]. Evidence suggests risk of NTD increases with folate insufficiency levels higher than ranges defining folate deficiency[34]. We dichotomized plasma folate levels based on the stringent threshold indicative of folate deficiency using a cut-off point of 4 ng/ml[15,16]. We fit univariate and multivariate binomial regression models to assess the degree to which each unit change in plasma folate led to a change in the risk of folate deficiency at six and twelve months, and obtained risk ratio estimates[35]. In multivariate models, we included potential confounders known to be associated with folic acid intake and/or plasma folate levels, or have been identified in regression models (p<0.2) to be significantly related to dietary intake of folic acid at baseline. Relative risks were adjusted for age (18 to <2 6, 26 to <36 and 36–49 years), years of formal education (0–7 years, 8–11 years and ≥12 years), occupation (business/professional, skilled formal, skilled informal, unskilled, unemployed), body mass index (<18.5, 18.5 to <25, 25 to <30, ≥30 kg/m2), household dietary diversity score (1–12), baseline intake of fortified wheat-based foods (servings per day), intake of vegetables and total energy intake (kcals/day).The number of household assets were computed from a simple count that included TV, radio, generator, fan, bike, car, couch, fridge, as well as access to electricity and potable water, allowing classification into 3 socioeconomic groups thus: 0–5, 6–8, and 9–10 [36]. Covariates with missing data were retained in the analysis using the missing indicator method[37].P-values were two-sided and significance was set at < 0.05. Final data set S2 File was compiled and all statistical analyses were conducted using SAS version 9.2 (SAS Institute Inc).