Riboflavin (vitamin B2) is required for many metabolic pathways, usually as a precursor of the flavin coenzyme, NAD, in redox reactions (McCormick, Reference McCormick and Dulbecco1997). Riboflavin deficiency is particularly common in regions where intakes of dairy products and meat are low (Neumann et al. Reference Neumann, Bwibo, Murphy, Sigman, Whaley, Allen, Guthrie, Weiss and Demment2003). Deficiency causes impaired growth, cheilosis, angular stomatitis, glossitis, dermatitis and impaired vision (Venkataswamy, Reference Venkataswamy1967; Batey et al. Reference Batey, Daneshgar and Eckhert1992; FAO/WHO, 2001). Schoolchildren, in both developing and industrialised countries, are an age group at high risk for riboflavin deficiency (Powers, Reference Powers2003).
In Western Africa, anaemia affects 50–80 % of the children (ACC/SCN, 2000; WHO/UNICEF/UNU, 2001). There are many aetiologies of anaemia in African children. These include malaria, blood loss from parasitic infections, vitamin A deficiency and diets low in bioavailable iron (ACC/SCN, 2000). Riboflavin deficiency may also impair erythropoiesis and contribute to anaemia (Foy & Kondi, Reference Foy and Kondi1953; Foy et al. Reference Foy, Kondi and Mbaya1964). Suggested mechanisms for this effect of riboflavin deficiency are decreased mobilisation of iron from stores (Sirivech et al. Reference Sirivech, Frieden and Osaki1974; Powers, Reference Powers1995), decreased iron absorption and increased iron losses (Adelekan & Thurnham, Reference Adelekan and Thurnham1986; Powers et al. Reference Powers, Wright and Fairweather-Tait1988; Butler & Topham, Reference Butler and Topham1993). Whereas these mechanisms have been investigated in animals, there are few data from human studies. Riboflavin supplements given concurrently with iron improve the response to iron supplementation in pregnant and lactating women, adult males and schoolchildren (Buzina et al. Reference Buzina, Jusic, Milanovic, Sapunar and Brubacher1979; Charoenlarp et al. Reference Charoenlarp, Pholpothi, Chatpunyaporn and Schelp1980; Ajayi et al. Reference Ajayi, Okike and Yusuf1990). A stable isotope study in adult Gambian men showed a greater increase in Hb values with riboflavin supplementation compared with controls (both groups received iron supplements), but failed to demonstrate an increase in iron absorption, suggesting that the improvement in Hb may have been due to iron mobilisation from stores (Fairweather-Tait et al. Reference Fairweather-Tait, Powers, Minski, Whitehead and Downes1992).
The aims of the present study were to: (1) determine the prevalence of riboflavin deficiency in school-age children in southern Côte d'Ivoire; (2) estimate the riboflavin content of the local diet; and (3) investigate if riboflavin status contributes to anaemia or iron deficiency in this population.
Subjects and methods
Study site
The study sites were two villages in Côte d'Ivoire from geographically distinct regions. Orbaff is a rural village in the Dabou district of Côte d'Ivoire, 10 km from the southern coast. The climate is tropical, with a temperature of ≥ 27°C and a relative humidity of ≥ 80 % much of the year. Plantain, rice, yam and dried, smoked fish are eaten regularly. Bringakro is further north (District Toumodi), about 180 km northwest of Abidjan. The region is situated in the transition zone from rainforest to savannah, with a mean temperature of 27°C and a relative humidity of 70 %. Staple foods in this area are yam, plantain, cassava and red palm oil. The surveys were conducted in May 2004 in Orbaff (Wegmuller et al. Reference Wegmuller, Camara, Zimmermann, Adou and Hurrell2006), and in May 2005 in Bringakro.
Subjects and enrolment
In both villages, a random sample was taken from children attending grades 1–5 in primary schools, with 50 and 28 % of children in these grades enrolled in Bringakro and Orbaff, respectively. Informed written consent was obtained from the school directors and the parents of the children. The Ethical Committee of the Swiss Federal Institute of Technology in Zürich, Switzerland, and the Scientific Council of the CSRS in Abidjan, Côte d'Ivoire gave ethical approval for the study. A total sample size of 280 children was calculated assuming a change in Hb of 3 g/l as clinically relevant, a standard deviation (sd) of Hb of 12 g/l, a significance level of 95 %, and a power of 75 % for a two-sided outcome.
Screening
Weight and height were measured upon registration (surname, first name, sex, date of birth) of the child. For these measurements, children removed their shoes and wore light clothing. Weight was recorded to the nearest 0·1 kg (Skyline, Soehnle-Waagen GmbH & Co. KG, Murrhardt, Germany), and height to the nearest 0·5 cm (Person check®, KaWe, Asperg, Germany) by the local nurse and an experienced technician. Age was obtained from the school registers, where age is taken from birth certificates. Venous blood samples (7·5 ml) were drawn in the morning into EDTA-treated evacuated tubes, and blood samples were kept in the dark on ice during transport to a nearby laboratory. A morning casual spot urine sample was collected for measurement of microhaematuria (as an indicator for bilharziosis). After the screening, children with iron deficiency were given supplemental oral iron for 12 weeks (daily 60 mg of iron as FeSO4), and children with bilharziosis were given a single dose of praziquantel (40 mg/kg).
Dietary assessment
Three-day weighed food records were collected in Orbaff in 24 randomly selected households. The records were done on three consecutive days (including two weekdays and one weekend day). In the village, family meals are traditionally eaten from shared bowls, but to estimate individual food intake during the 3 days of recording, families ate from individual bowls. Edible portions of all food and beverages were weighed using a Soehnle scale (Vera 67 002, Soehnle-Waagen GmbH & Co. KG, Murrhardt, Germany) accurate to ± 1 g during preparation and consumption. Meals typically consisted of a cooked staple and a sauce. When the meal was ready, the total weight of the cooked staple and sauce was weighed. After the meal, the food that remained uneaten was weighed and the proportion of staple to sauce was estimated by a team of experienced investigators. Subjects absent during the day were asked about their food intake and amounts eaten outside the home; these estimations were then added to the food records. Riboflavin intake was calculated using food composition tables (Toury et al. Reference Toury, Giorgi, Favier and Savina1965; FAO, 1968; Bundesinstitut für gesundheitlichen Verbraucherschutz und Veterinärmedizin, 1996) and food analysis software EBISpro (University of Hohenheim/Stuttgart, Germany).
Laboratory analysis
In Orbaff, Hb measurement was performed on the day of blood sampling on an AcT8 Counter (Beckman Coulter, Krefeld, Germany) using three-level controls provided by the manufacturer. In Bringakro, because the AcT8 Counter was not available, the Hemocue system was used (Hemocue, Angelsborg, Sweden) using two-level controls provided by the manufacturer. Venous blood was used for both methods; values obtained by these two Hb methods from venous blood are highly correlated (Morris et al. Reference Morris, Pont and Lewis2001). Zinc protoporphyrin (ZPP) was measured on washed red blood cells within 7 d of blood sampling in duplicate with a haematofluorometer (Aviv Biomedical, Lakewood, NJ, USA) (Hastka et al. Reference Hastka, Lasserre, Schwarzbeck, Strauch and Hehlmann1992), with standards provided by the manufacturer. C-reactive protein (CRP) in samples from Bringakro was measured using an automated chemiluminescent high sensitivity immunoassay system (IMMULITE®, Diagnostic Products Corporation, Los Angeles, CA, USA), and Orbaff CRP samples were measured using nephelometry (TURBOX®, Orion Diagnostics, Espoo, Finland). Anaemia was defined as Hb < 120 g/l in children aged ≥ 12 years, and Hb < 115 g/l in children aged 5–12 years (WHO/UNICEF/UNU, 2001). Iron deficiency was defined by a ZPP value >40 μmol/mol haem (Hastka et al. Reference Hastka, Lasserre, Schwarzbeck, Strauch and Hehlmann1992). CRP values < 10 mg/l were defined as normal (Pepys, Reference Pepys1981). Serum retinol (SR) was measured by HPLC (Merck-Hitachi, Tokyo, Japan) according to Tanumihardjo et al. (Reference Tanumihardjo, Permaesih and Muhilal2004) with reference material from the National Institute of Standards and Technology (Gaithersburg, MD, USA); vitamin A deficiency was defined as an SR < 0·70 μmol/, and low vitamin A status as an SR < 1·05 μmol/l (WHO, 1996). Riboflavin was measured by the erythrocyte glutathione reductase activity coefficient (EGRAC) assay using a modification of the method of Dror et al. (Reference Dror, Stern and Komarnitsky1994). In our laboratory, the inter- and intra-assay CV of the EGRAC method is 3 and 4 %, respectively. Although there is no consensus for EGRAC cut-off values for riboflavin deficiency (Tillotson & Baker, Reference Tillotson and Baker1972; Prasad et al. Reference Prasad, Lakshmi and Bamji1992; Sauberlich, Reference Sauberlich1999), we used cut-offs specified for this EGRAC assay (Dror et al. Reference Dror, Stern and Komarnitsky1994), i.e. values >1·2 indicated marginal deficiency and values >1·4 indicated clear deficiency (Sauberlich, Reference Sauberlich1999). For statistical analysis, EGRAC values >1·2 were used. On the day of blood sampling, whole blood was used to prepare a thick and a thin smear for malaria parasites by the Giemsa coloration technique (WHO, 1990). Parasites were counted against leucocytes; fields containing >200 leucocytes were counted and if < 10 parasites were identified, the counting continued up to 500 leucocytes. These counts were converted to the number of parasites per μl of blood, assuming a leucocyte count of 8000/μl (WHO, 1990). On the day of blood sampling, haematuria, as an indirect marker for infection with Schistosoma haematobium, was assessed semiquantitatively by dipstick (Combur 7, Roche Diagnostics GmbH, Mannheim, Germany).
Statistical analysis
Data processing and statistical analysis were done using Excel (XP 2002; Microsoft, Redmond, CA, USA) and SPSS 13.0 (SPSS Inc., Chicago, IL, USA). Height-for-age and weight-for-age Z-scores were calculated using Epi-Info (Epi-Info Version 3.3.2, Centers for Disease Control and Prevention, Atlanta, GA, USA). All data except ZPP were normally distributed and are presented as means and their sd. ZPP data were skewed and are presented as medians (ranges); for statistical analysis, these values were log-transformed. One-way ANOVA was used to test for differences for all scaled variables among the different geographic areas; the Mann–Whitney test was applied for ordinal data. Two-tailed Pearson's correlation coefficients were used to examine the relationship between iron status or anaemia and riboflavin status, malaria, inflammation, SR and haematuria. Stepwise linear regression on Hb and logZPP was performed with age, sex, EGRAC, SR, CRP, malaria and Hb or logZPP as independent variables. Odds ratios for the risk of anaemia or iron deficiency in riboflavin deficiency were calculated. P values < 0·05 were considered statistically significant.
Results
Three-day weighed food records
The 3 d weighed food records were kept in 24 families comprising 207 subjects (median age: 16 years; range: 2–81 years); the results for children 6–15 years of age (n 71) are reported here. Daily intakes of riboflavin ranged from 0·17 to 1·16 mg, with a median of 0·42 mg, representing 70 and 47 % of the estimated average requirement for children 6–8 years of age and those aged 9–13, respectively (Institute of Medicine, 2001). Riboflavin intake was low due to infrequent consumption of animal foods, with the exception of smoked fish, and the negligible riboflavin content of cassava, the dietary staple. Median (range) iron intake was 9·8 mg (3·6–19·8) mg/d, well above the estimated average requirement of 6·3 mg/d for 6- to 15-year-old children. Only 21 % of children had iron intakes below the estimated average requirement. However, the bioavailability of the dietary iron was probably low, as 90 % of iron was non-haem iron.
Anthropometry
Mean age of the screened children was 10·2 (sd 2·3) years, mean weight and height were 31·8 (sd 10·1) kg and 1·38 (sd 0·15) m, respectively. Forty-four percent of the children were female. As shown in Table 1, the children from the two villages differed significantly in sex ratio, age, weight and height. However, height-for-age as well as weight-for-age Z-scores were not significantly different. Eight percent of the children were stunted (height-for-age Z-score less than − 2 sd) and 11 % were undernourished (weight-for-age Z-score less than − 2 sd).
Table 1 Anthropometric, nutritional and health characteristics of Ivorian children in two villages of Côte d'Ivoire: Orbaff, in the southern tropical zone, and Bringakro, in the savannah transition zone
EGRAC, erythrocyte glutathione reductase activity coefficient; CRP, C-reactive protein.
* Mean and standard deviation.
† Schools differ significantly by ANOVA (P < 0·05).
‡ Median (range).
§ n 162.
‖ n 55.
¶ n 107.
** n 159.
†† n 49.
‡‡ n 161.
§§ n 51.
Prevalence of anaemia, iron deficiency and iron deficiency anaemia
A high prevalence of anaemia (52 %) was found in the studied population (Table 1). Nearly 60 % of all children had iron deficiency, and iron deficiency was present in 69 % of the anaemic children. Iron deficiency anaemia was present in 36 % of all children. As expected, ZPP and Hb were significantly correlated (Table 2), but no other parameter alone influenced these two variables. Vitamin A deficiency is thought to contribute to anaemia (Reddy, Reference Reddy1998), but only 2 % of the children in the sample were vitamin A deficient. Comparing riboflavin-deficient (n 182) and riboflavin-sufficient (n 99) children, ZPP concentrations were 51 (17–296) and 46 (20–49) μmol/mol haem, the prevalence of iron deficiency was 63·2 and 52·5 %, and the prevalence of iron deficiency anaemia was 35·7 and 36·4 %, respectively.
Table 2 Pearson's correlation coefficients between nutrition and health-related variables in school-aged children in two villages of Côte d'Ivoire
ZPP, zinc protoporphyrin; CRP, C-reactive protein; EGRAC, erythrocyte glutathione reductase activity coefficient; SR, serum retinol.
* Correlation is significant at the 0·05 level.
† Correlation is significant at the 0·01 level.
Prevalence of malaria parasites and microhaematuria
Despite a high prevalence of malaria parasites (49 %), the severity of infestation was low, with only 8 % of the children having a parasite load >1000/μl blood, and only two children having clinical malaria, as defined as a parasite load >5000/μl blood (Trape, Reference Trape1985). Plasmodium load was positively correlated with SR, but with no other variable (Table 2). Microhaematuria, an indicator for S. haematobium infection, was present in 19 % of the children.
Prevalence of inflammation
The prevalence of inflammation or infection, as indicated by elevated CRP concentrations, was moderate in both villages. Only about 15 % of the children had an elevated CRP despite a considerably higher prevalence of Plasmodium parasitaemia. This discrepancy may at least partly be explained by the generally low malaria severity.
Prevalence of riboflavin deficiency
Two-thirds of the children suffered from riboflavin deficiency, but the degree of deficiency was mild. Only 2 % of children had EGRAC values >1·4, suggesting clear deficiency. Of the riboflavin-deficient children, 7·5 % showed EGRAC values >1·4. EGRAC was not correlated with any other measured variable (Table 2).
Regression analysis and odds ratios
In the stepwise linear regression on Hb, age (P < 0·001), log ZPP (P < 0·001) and CRP (P < 0·05) were significant predictors. With logZPP as the dependent variable, Hb (P < 0·001) and EGRAC (P < 0·05) were significant predictors. Overall, correlations were modest.
In riboflavin-deficient subjects, the odds ratios (95 % CI) for anaemia and iron deficiency were 0·69 (0·42, 1·13) and 1·55 (0·94, 2·55), respectively. Because EGRAC values may be influenced by malaria (see below), odds ratios were also calculated after excluding subjects with malaria. In riboflavin-deficient subjects (EGRAC >1·2) without Plasmodium infection (n 80), the odds ratio for iron deficiency was significant (3·07; 1·12, 8·41).
Discussion
Mild riboflavin deficiency was very common in the study children, with nearly two-thirds of the children affected. A previous Ivorian study reported EGRAC values >1·68, but no information on the age of the subjects or the prevalence of riboflavin deficiency was provided (Arnaud et al. Reference Arnaud, Malvy, Richard, Faure and Chaventre2001). A food survey in central Côte d'Ivoire found low consumption of dairy products and meat, the usual major dietary sources of riboflavin, but riboflavin intakes were not reported (Staubli-Asobayire, Reference Staubli-Asobayire2000). Although there are few data on the riboflavin status of children in sub-Saharan Africa, the few published studies suggest that riboflavin deficiency may be common. A study in Botswana reported that 33–40 % of the examined children had an EGRAC ≥ 1·4 (Abrams et al. Reference Abrams, Mushi, Hilmers, Griffin, Davila and Allen2003). In Kenya, approximately one-third of the children were riboflavin deficient, measured as red blood cell riboflavin (Siekmann et al. Reference Siekmann, Allen, Bwibo, Demment, Murphy and Neumann2003). In 2- to 5-year-old South African children, intakes of riboflavin were only 49 % of the RDA (Faber et al. Reference Faber, Jogessar and Benade2001).
In this study, although malaria was endemic, with 49 % of the children infected, the parasite load was generally low. This prevalence is similar to that reported in a previous Ivorian study, but the severity of infection was higher in that study (Staubli Asobayire et al. Reference Staubli Asobayire, Adou, Davidsson, Cook and Hurrell2001). A study conducted in eastern Côte d'Ivoire found that malaria affected 74–86 % of 7- to 11-year-old children and 68 % of children >12 years old (Girardin et al. Reference Girardin, Dao and Koudou2004). Microhaematuria as an indicator of S. haematobium infection varied markedly between villages, with nearly one-third of the children affected in Bringakro. This high prevalence of S. haematobium infection is consistent with a previous report from this region of Côte d'Ivoire, where 24·6 % of the children had microhaematuria, although data varied markedly depending on the nearby presence of standing water (N'Goran et al. Reference N'Goran, Utzinger, Traore, Lengeler and Tanner1998). Anaemia prevalence in the present study (52 %) was similar to that found in an earlier report in Ivorian children (46 %) (Staubli Asobayire et al. Reference Staubli Asobayire, Adou, Davidsson, Cook and Hurrell2001). However, in that study, the prevalence of elevated ZPP levels was higher (83 %) than in the present study (59 %).
There are several limitations to this study. First, the use of a single measure of iron status (ZPP) probably reduced our ability to define iron deficiency accurately (Zimmermann et al. Reference Zimmermann, Molinari, Staubli-Asobayire, Hess, Chaouki, Adou and Hurrell2005). Previous studies have suggested that ZPP may be elevated by infection and/or inflammation, and thus iron deficiency using ZPP may be overestimated in areas of endemic infection (Staubli Asobayire et al. Reference Staubli Asobayire, Adou, Davidsson, Cook and Hurrell2001). However, in the present study, there was no correlation between ZPP and CRP, presence of malaria or suspected schistosomiasis. It has been suggested that acute malaria may spuriously elevate EGRAC (Anderson et al. Reference Anderson, Scattoni, Perry, Galvan, Giuberti, Buonocore and Vullo1994; Traunmuller et al. Reference Traunmuller, Ramharter, Lagler, Thalhammer, Kremsner, Graninger and Winkler2003). However, acute malaria was rare in our study population, and there was no relationship between EGRAC and malaria, so our estimate of riboflavin deficiency based on EGRAC is likely to be valid.
The public health impact of widespread riboflavin deficiency in schoolchildren is unclear; but it may influence iron metabolism (Sirivech et al. Reference Sirivech, Frieden and Osaki1974; Adelekan & Thurnham, Reference Adelekan and Thurnham1986; Powers et al. Reference Powers, Wright and Fairweather-Tait1988; Butler & Topham, Reference Butler and Topham1993; Powers, Reference Powers1995). In our sample, anaemia and iron deficiency were common, with prevalences >50 %, and two-thirds of the anaemic children were iron deficient. Neither linear regression nor calculation of relative risk identified riboflavin deficiency as a significant predictor of Hb or anaemia. Thus, our data do not support a detrimental effect of riboflavin deficiency on anaemia, as suggested by earlier studies (Foy & Kondi, Reference Foy and Kondi1953, Reference Foy and Kondi1958), a finding that might be explained by the low severity of the riboflavin deficiency in our sample. However, the risk of iron deficiency was increased 3-fold in children with riboflavin deficiency and without malaria. There are two likely explanations for this association. First, poor overall quality of the diet may have led to multiple micronutrient deficiencies; however, there was a very low prevalence of vitamin A deficiency, and iodine status appeared to be adequate. Secondly, the results could reflect impairment of iron absorption and/or metabolism by riboflavin deficiency, as suggested by earlier studies (Powers et al. Reference Powers, Wright and Fairweather-Tait1988; Butler & Topham, Reference Butler and Topham1993; Powers, Reference Powers1995). In three trials in children and pregnant women (Charoenlarp et al. Reference Charoenlarp, Pholpothi, Chatpunyaporn and Schelp1980; Powers et al. Reference Powers, Bates, Prentice, Lamb, Jepson and Bowman1983; Suprapto et al. Reference Suprapto, Widardo and Suhanantyo2002), riboflavin and iron supplementation produced a greater increase in Hb compared with iron supplements given alone, although the results may have been confounded by concomitant folic acid supplementation in two (Charoenlarp et al. Reference Charoenlarp, Pholpothi, Chatpunyaporn and Schelp1980; Suprapto et al. Reference Suprapto, Widardo and Suhanantyo2002). In contrast, another trial reported no additional benefit of riboflavin plus iron compared with iron supplementation alone in young adults (Powers & Bates, Reference Powers and Bates1987). A study in a small number of pregnant women reported that riboflavin supplementation alone or in combination with iron supplementation did not prevent a decrease in Hb, although the decrease was less in the treated group than in the control group (Powers et al. Reference Powers, Bates and Lamb1985). Similarly, no effect could be found in a riboflavin supplementation trial in Croatian schoolchildren with adequate Hb levels (Suboticanec et al. Reference Suboticanec, Stavljenic, Schalch and Buzina1990). Taken together, these data suggest that the effect of riboflavin status on Hb is variable, and may be confounded by the multifactorial aetiology of anaemia, particularly in countries in sub-Saharan Africa.
In conclusion, there is a high prevalence of mild riboflavin deficiency in Ivorian schoolchildren, but severe deficiency was rare. Riboflavin deficiency was associated with iron deficiency in children free of malaria parasites. However, there was no significant association between riboflavin status and anaemia.
Acknowledgements
We thank the participating children, school directors and teachers, and we thank the authorities of the villages and health districts. We are grateful for the assistance of Jean-Paul Gnanzou, Yao Edouard Koffi from the Institut National de Santé Publique; Traoré Mahamadou and Kigbafori Silué from the Centre Suisse de Recherches Scientifiques; and Christophe Zeder, Renata Wälchli, Valeria Galetti and Ralf Biebinger from ETH Zurich. We thank Dr Paul Lohmann GmbH KG (Emmerthal, Germany) who supplied the iron sulphate tablets. We thank Luciano Molinari from the Children's Hospital in Zurich for support in statistical analysis. Financial support for the study was from the ETH through a private donation. None of the authors has financial or personal conflicts of interest with regard to this study.
