European Journal of Clinical Nutrition (2002) 56, 666–673 ß 2002 Nature Publishing Group All rights reserved 0954–3007/02 $25.00 www.nature.com/ejcn ORIGINAL COMMUNICATION Effects on serum retinol of multi-micronutrient supplementation and multi-helminth chemotherapy: a randomised, controlled trial in Kenyan school children

D Mwaniki1, B Omondi1, E Muniu1, F Thiong’o2, J Ouma2, P Magnussen3, PW Geissler3, K Fleischer Michaelsen4 and H Friis3,4*

1Centre for Public Health Research, Kenya Medical Research Institute, Nairobi, Kenya; 2Division of Vector Borne Diseases, Ministry of Health, Nairobi, Kenya; 3Danish Bilharziasis Laboratory, Charlottenlund, Denmark; and 4Research Department of Human Nutrition, The Royal Veterinary and Agricultural University, Frederiksberg, Denmark

Objective: To assess the effects of multi-micronutrient supplementation and multi-helminth chemotherapy on serum retinol concentration, using schools as a health delivery system. Study area and population: From 19 primary schools in Bondo District, western Kenya, 977 children between 9 and 18 y were included in the trial. The 644 (65.9%) children on whom baseline serum retinol was available were included in this study. Design: A randomised, placebo-controlled, double-blind, two-by-two factorial trial on the effects of multi-micronutrient supplementation and multi-helminth chemotherapy on serum retinol after 8 months. Intervention: Single treatment with albendazole (600 mg) and praziquantel (40 mg=kg of body weight) and daily multi- micronutrient supplementation with tablet containing 1000 mg vitamin A. Results: Micronutrient supplementation (0.08 mmol=l, 95% CI 0.01, 0.14; P ¼ 0.025), but not treatment (0.03 mmol=l, 95% CI 7 0.04, 0.10; P ¼ 0.38), increased serum retinol. However, treatment did increase serum retinol in S. mansoni-infected (0.09, 95% CI 0.02, 0.16; P ¼ 0.009), but not in uninfected children ( 7 0.07, 95% CI 7 0.18, 0.03; P ¼ 0.18; interaction, P ¼ 0.01). Similarly, reduction in egg output of S. mansoni, but none of the geohelminth, was a predictor, corresponding to a 0.008 mmol=l (95% CI 0.00002, 0.02; P ¼ 0.049) increase in serum retinol per 100 epg reduction. Interestingly, interactions were found between age and sex (P ¼ 0.046), and malaria parasitaemia and sickle cell phenotype (P ¼ 0.04). Conclusion: Multi-micronutrient supplementation and reduction in S. mansoni egg output increased serum retinol, irrespective of initial serum retinol. Sponsorship: The Danish International Development Assistance. European Journal of Clinical Nutrition (2002) 56, 666 – 673. doi:10.1038=sj.ejcn.1601376

Keywords: micronutrients; anthelminthic treatment; malaria; S. mansoni; hookworm; T. trichiura; A. lumbricoides; vitamin A; school children; Kenya

Introduction Vitamin A deficiency is widespread in developing countries, *Correspondence: H Friis, Research Department of Human Nutrition, The particularly in young children and women where it is an Royal Veterinary and Agricultural University, Rolighedsvej 30, DK-1958 important cause of immune function impairment (Semba, Frederiksberg C, Denmark. 1994) and increased risk of morbidity and mortality from E-mail: [email protected] infectious and other diseases (Fawzi et al, 1993; Sommer et al, Guarantor: H Friis. Contributors: DM, EM, FT, JO, PM, PWG, KFM and HF were responsible 1984; West et al, 1999). The main causes of vitamin A for concept and planning. DM, BO, JO, FT and PWG were responsible deficiency are inadequate intake of vitamin A-rich foods for laboratory and field work. DM, EM and HF were responsible for such as liver, full milk, butter, eggs, whole fish, green leafy data processing. DM, PM, KFM and HF were responsible for writing vegetables and yellow-orange fruits, as well as increased the manuscript. Received 8 June 2001; revised 24 October 2001; requirements due to a high burden of infectious diseases. accepted 29 October 2001 Accordingly, rational long-term measures to increase vitamin Micronutrients, anthelminthics and serum retinol D Mwaniki et al 667 A status of these populations should comprise promotion of dently randomised to multi-helminth chemotherapy or food diversification and prevention of infectious diseases. identical-looking placebo. Micronutrient supplementation Nevertheless, in the meantime, short-term measures such as comprised a tablet with 13 vitamins and minerals (vitamin vitamin A supplementation and treatment of prevalent A 1000 mg, C 50 mg, B1 1.4 mg, B2 1.6 mg, B6 1.7 mg, B12 2 mg, chronic infections could target high-risk groups where fea- folate 150 mg, niacin 16 mg, iron 18 mg, zinc 20 mg, copper sible and cost-effective. For example, vitamin A deficiency is 2.0 mg, iodine 150 mg, selenium 40 mg; Almega, Ringsted, also frequent in school-age children (Friis et al, 1997), and Denmark) on all schooldays for a full school year. Multi- schools offer a useful entry point for programmes that could helminth chemotherapy was given to children found comprise micronutrient supplementation, treatment of para- infected with any of the geohelminths or S. mansoni with sitic infections, school gardens and health education. albendazole and praziquantel, respectively. Albendazole was In western Kenya, malaria is holoendemic, Schistosoma (S.) given as a single dose of 600 mg, and praziquantel as a single mansoni, hookworm (Necator americanus), Trichuris (T.) tri- dose of 40 mg=kg of body weight. The examinations chiura and Ascaris (A.) lumbricoides infections are prevalent, described below were done at baseline and repeated after 8 and vitamin A status generally poor. We conducted a study months. on the effects of multi-micronutrient supplementation and multi-helminth chemotherapy on serum retinol concentra- tion (serum retinol) among 9 – 18-y-old school children in 19 Serum retinol and haematology primary schools in western Kenya, with the aim to assess the Blood samples were taken from the antecubital veins role of these interventions in future school-based health between 9 am and 1 pm. Serum was prepared and kept programmes. An inverse association between S. mansoni frozen at 7 20C for less than 3 months, and serum retinol egg output, but neither geohelminth egg output nor malaria was measured by high performance liquid chromatography parasitaemia, and low serum retinol, from baseline data has (Hitachi Ltd, Japan) as previously described (Friis et al, 1997). previously been reported (Friis et al, 1997). In this paper we To 100 ml serum were added 5 ml retinyl acetate (100 mg=ml) as report the effect of the interventions on serum retinol. internal standard. The proteins were precipitated using 100 ml of methanol, and retinol was extracted using 200 ml of diethyl ether : dichlormethane : isopropanol mixture Subjects and methods (80 : 19 : 1). The extract was reconstituted with 100 ml The study was conducted on the shores of Lake Victoria in mobile phase (methanol : water, 95 : 5, v=v) and 20 ml injected Bondo District, Nyanza Province in western Kenya. The into a guard-fitted normal-phase stainless steel column residents of the study area were members of the Luo com- (microbondapak C , 3.9Â300 mm, particle size 10 mm, munity and their principal occupations were subsistence 18 Waters Assoc, Milford, MA, USA). Inter-batch coefficient of crop farming, raising Zebu cattle, commercial fishing and variation was 6.2%. A serum retinol value of 0.70 mmol=l petty trade. Malaria was known to be holoendemic in the defined low and 1.05 mmol=ldefined marginal levels. Hemo- study area, and intestinal helminths and S. mansoni were globin concentration (Hb) and total white blood cell (WBC) prevalent, whereas S. haematobium was not endemic. counts were determined using an electronic counter (M530, The study was a randomised, placebo-controlled, double- Coulter Electronics Ltd, England), and differential WBC blind trial on the effects of daily multi-micronutrient sup- counts were determined using conventional manual plementation and multi-helminth chemotherapy, using a methods. two-by-two factorial design. From grades 5 and 6 in 19 primary schools in the study area, 1096 children were iden- tified who had their parents consent to participate. Of these, 38 were ill and therefore received initial treatment and were Clinical examination and anthropometry referred for more definitive management and follow-up. The children were examined clinically, blood samples were These and 81 children that were unavailable for baseline collected, and a thin and a thick blood slide were prepared. examinations were excluded. The remaining 977 children Additionally, questionnaire data and anthropometric mea- were included in the study. Despite initial consent, some surements were obtained on the day of blood sampling. parents or children did not accept blood sampling. Baseline Finally, stool samples were collected on different days. data on serum retinol was therefore only available on 644 Height and weight were measured with the children bare- (65.9%) of the 977 children, and the results of the 8 months footed and wearing light standard clothing. Heights were follow-up of these children are reported here. measured to the nearest 0.5 cm and weights to the nearest 0.1 kg, using a scale-stadiometer (Seca, Germany). Height and weight were related to references as standard deviation Study interventions scores (Z-scores). Height-for-age (HAZ) and weight-for-age After completion of the baseline examinations described (WAZ) Z-scores were computed based on the NCHS-WHO below, the children were randomised to micronutrient sup- growth reference curves. HAZ and WAZ below 7 2 were plementation or identical-looking placebo and indepen- considered to reflect stunting and underweight, respectively.

European Journal of Clinical Nutrition Micronutrients, anthelminthics and serum retinol D Mwaniki et al 668 Infections Review Committee, and the Ministry of Health, Kenya. The A single stool sample was collected on each of two or three children and their parents were given information about the consecutive days around the time of blood sampling, and study, and the parents gave their written consent. The examined quantitatively for intestinal helminths and schis- Danish Central Medical Ethics Committee also approved tosome eggs using the World Health Organization (WHO) the study. All children found to be infected were treated at modification of the Kato – Katz method (World Health Orga- the end of the study. nization, 1991). In short, a duplicate of 50 mg cellophane faecal thick smears was prepared from each of the stool samples. The smears were examined for hookworm eggs Results within 1 h, and after at least 24 h clearing for eggs of S. The 644 children included in this study were similar to those mansoni, T. trichiura and A. lumbricoides. The egg output of the 333 children on whom baseline serum retinol data was was expressed as mean eggs per gram faeces (epg). From all not available, with respect to all other baseline characteris- children a thick blood slide was examined for malaria para- tics, such as age, sex, nutritional status and prevalence and sites. The number of parasites per 200 WBC was counted. If intensity of infections (data not shown). The 644 children below 10 parasites, then 500 WBC were examined. The had a mean age of 13.0 (range 9 – 18) y, and 339 (52.6%) were number of malaria parasites per ml of blood was computed girls and 305 (47.4%) boys. The mean weight- and height- based on the total WBC counts. Questionnaire morbidity for-age Z-scores were 7 1.11 (95% confidence interval (CI) data, elevated body temperature and neutrophilia were used 7 1.17, 7 1.05) and 7 0.96 (95% CI 7 1.04, 7 0.89), as indicators of infection. Structured interview data on fever, respectively. Nutritional status was lower in boys than in malaria and diarrhoea within the preceding 2 weeks were girls, as assessed by weight-for-age ( 7 1.24 vs 7 0.99, used as clinical indicators of infection. Children rarely P ¼ 0.00002) and height-for-age ( 7 1.12 vs 7 0.81, reported fever, as fever was apparently perceived as malaria P ¼ 0.00003) Z-scores. The mean serum retinol was 0.94 or headache. Neutrophilia was arbitrarily defined as an (95% CI 0.91, 0.97) mmol=l, and lower for boys than for absolute neutrophil count above the 87.5 percentile. Sub- girls (0.88 vs 0.99 mmol=l, P ¼ 0.00008), as previously lingual temperatures were taken using a digital thermometer reported (Friis et al, 1997). The prevalence of malaria para- (Becton Dickinson, NJ, USA), and > 37.0Cdefined elevated sitaemia was 60%, whereas it was 14% for A. lumbricoides, body temperature. 45% for T. trichiura, 54% for hookworm and 71% for S. mansoni. Boys had lower prevalence of Trichuris infection Statistical analysis (41 vs 49%, P ¼ 0.032), and marginally significantly higher The distribution of serum retinol was assessed using rankit- prevalence of hookworm (58 vs 51%, P ¼ 0.09), whereas there plots. The two-sample t-test and the chi-square test were used was no sex difference in ascaris, S. mansoni or malaria infec- to test for differences in means and proportions, respectively. tion (data not shown). The geometric mean T. trichiura egg If no interaction was found between the two interventions, output was lower in boys than in girls (35 vs 56, P ¼ 0.01), but then it was allowed to compare the active vs the placebo there were no sex differences in intensity of malaria, ascaris, groups for each intervention without taking the other inter- S. mansoni and hookworm infections (data not shown). vention into consideration. Multiple regression analyses was employed to identify and estimate effects of predictors of serum retinol. The variables assessed were the two interven- Interventions and other predictors of change in serum tions (both coded 0 for placebo and 1 for the active com- retinol pounds), age, sex, HAZ, WAZ, hepato- and organomegaly, The simple randomisation used resulted in baseline equiva- malaria parasitaemia and helminth egg output, sickle cell lence between the intervention groups, except with respect phenotype and, as indicators of infection, elevated tempera- to organomegaly (Table 1). After 8 months follow-up, 496 ture, neutrophilia and reported morbidity. If interactions (77.0%) of the 644 children were followed-up and their were identified, then interaction terms were computed to serum retinol determined. These children were not different allow estimation of the effect of one of the variables sepa- from the 148 (23.0%) children not followed-up, with respect rately for each level of the effect modifying variable. Data on to baseline characteristics, such as age, sex, nutritional status malaria parasitemia, S. mansoni and intestinal helminths egg and prevalence and intensity of infections (data not shown).

output were transformed, using a log10(x þ 1) transforma- Because there was no interaction between micronutrient tion. Residual analysis was performed by plotting standar- supplementation and anthelminthic treatment as assessed dized residuals against predicted values and continuous in multiple regression analyses (P ¼ 0.37), the effect of one independent variables. The levels of significance used was intervention could be assessed without taking the other 0.05. intervention into consideration. Serum retinol increased from baseline to 8 months in all groups (Table 2), but the Permissions mean increase was 0.08 mmol=l (95% CI 0.01, 0.14) larger in Permission to conduct the study was obtained from the children assigned to multi-micronutrient supplementation Kenya Medical Research Institute and National Ethical compared with placebo (0.31 vs 0.23 mmol=l, P ¼ 0.025). In

European Journal of Clinical Nutrition Micronutrients, anthelminthics and serum retinol D Mwaniki et al 669 Table 1 Baseline characteristics of 644 children independently randomised to daily supplementation with multi-micronutrients or placebo, and treatment with anthelminthics or placebo

Placebo, placebo Micronutrients, placebo Placebo, anthelminthics Micronutrients, anthelminthics (n ¼ 163) (n ¼ 166) (n ¼ 157) (n ¼ 158)

Age in yearsa 12.9 (9.5, 16.5) 12.9 (9.0, 17.0) 13.0 (9.5, 17.0) 13.0 (9.2, 18.0) Sex (boys, %) 43 49 51 46

Nutritional status Weight-age Z-scoreb 7 1.10 ( 7 1.21, 7 0.99) 7 1.13 ( 7 1.25, 7 1.01) 7 1.09 ( 7 1.21, 7 0.97) 7 1.12 ( 7 1.24, 7 1.00) Height-age Z-scoreb 7 0.96 ( 7 1.11, 7 0.81) 7 1.01 ( 7 1.16, 7 0.86) 7 0.83 ( 7 1.00, 7 0.90) 7 1.05 ( 7 1.20, 7 0.90) Hemoglobin (g=l)b 123.5 (121.6, 125.5) 123.1 (121.0, 125.2) 125.0 (123.0, 127.0) 123.1 (121.1, 125.2) Serum retinol (mmol=l)b 0.94 (0.88, 0.99) 0.92 (0.87, 0.97) 0.93 (0.88, 0.99) 0.98 (0.93, 1.07)

Infections Malaria parasitaemiac 66% (495) 61% (550) 55% (537) 57% (589) S. mansonid 74% (89) 69% (62) 72% (64) 71% (68) Hookwormd 51% (56) 60% (40) 52% (60) 54% (47) T. trichiurad 49% (41) 43% (44) 48% (51) 42% (51) A. lumbricoidesd 14% (1349) 13% (1230) 15% (1585) 14% (599)

Organomegaly Hepatomegaly (%) 9 5 7 6 Splenomegaly (%) 16 9 5 7 aMean (range). bMean (95% confidence interval). cPrevalence (geometric mean malaria parasites per ml blood in infected children). dPrevalence (geometric mean eggs per gram (epg) faeces in infected, children).

Table 2 Effects on serum retinol (mean (95% confidence interval)) of daily multi-micronutrient supplementation and anthelminthic treatment among 496 children followed-up after 8 months

Supplementation (n ¼ 496) Treatment (n ¼ 496)

Placebo Micronutrient Placebo Anthelminthics Serum retinol (mmol=l) (n ¼ 238) (n ¼ 258) (n ¼ 257) (n ¼ 239)

Baseline 0.92 (0.87, 0.97) 0.95 (0.91, 0.99) 0.93 (0.89, 0.98) 0.95 (0.91, 0.99) 8 months 1.15 (1.10, 1.20) 1.26 (1.21, 1.31) 1.19 (1.14, 1.23) 1.23 (1.18, 1.28) Increase from baseline 0.23 (0.18, 0.27) 0.31 (0.26, 0.36) 0.26 (0.21, 0.31) 0.29 (0.24, 0.34) Difference between groups 0.08 (0.01, 0.14), P ¼ 0.025 0.03 ( 7 0.04, 0.10), P ¼ 0.38

contrast, the mean increase in serum retinol was not sig- Table 3 Final regression model with predictors of serum retinol at nificantly different (0.03 mmol=l, 95% CI 7 0.04, 0.10) in follow-up while controlling for baseline serum retinol, with regression children treated with anthelminthics compared to placebo coefficients (B), 95% confidence intervals (CI) and corresponding P- (0.29 vs 0.26 mmol=l, P ¼ 0.38). The mean increase in serum values; n ¼ 496, adjusted r2 ¼ 0.31 retinol in children assigned to placebo supplementation and Variables B 95% CI P placebo treatment was 0.20 mmol=l (95% CI 0.14, 0.27; P < 0.00001) from a mean of 0.92 mmol=l. These results Micronutrientsa 0.08 0.02, 0.14 0.008 b were confirmed in multiple linear regression analyses, Sex 7 0.11 7 0.17, 7 0.05 0.0003 Age (y) while controlling for baseline serum retinol. The final in girls 0.06* 0.03, 0.09 0.00005 model is shown in Table 3. Supplementation was a predictor in boys 0.02* 7 0.01, 0.05 0.15 of serum retinol at the 8 months follow-up examination S. mansoni infection 7 0.17 7 0.26, 7 0.08 0.0002 (0.09, 95% CI 0.02; 0.14, P ¼ 0.008), whereas treatment was Anthelminthics to S. mansoni infected 0.09** 0.02, 0.16 0.009 ¼ not (0.04, 95% CI 7 0.02, 0.10, P 0.18). There were no to S. mansoni uninfected 7 0.07** 7 0.18, 0.03 0.18 interactions between any of the interventions and baseline a serum retinol or low serum retinol (P > 0.10). An interaction Coded as 0 ¼ placebo and 1 ¼ active compound. b ¼ ¼ between age and sex was found (P ¼ 0.046). The effect of age Sex coded as 0 girls and 1 boys. The sex difference was estimated at the mean age of 13 y. was therefore estimated separately for boys and girls. The Regression coefficient with same superscript significantly different: *P ¼ 0.046; interaction was due to an increase in serum retinol with age **P ¼ 0.01.

European Journal of Clinical Nutrition Micronutrients, anthelminthics and serum retinol D Mwaniki et al 670 in girls (0.06, 95% CI 0.03, 0.09, P ¼ 0.00005), but not in 7 0.15, 0.11, P ¼ 0.81) malaria parasitaemia. In contrast, boys (0.02, 95% CI 7 0.01, 0.05, P ¼ 0.15). The regression there was no interaction between supplementation and coefficient of age in girls corresponded to a 0.06 increase in malaria in children allocated to active treatment (P ¼ 0.87). serum retinol with each year increase in age. Due to the Similarly, a marginal significant interaction between supple- interaction, the difference between boys and girls depends mentation and hepatomegaly (P ¼ 0.09) was found, but this on age. In the model, the effect of sex was estimated at a was similar in children allocated to placebo and active mean age of 13 y. With sex coded as 0 for girls and 1 for boys, treatment. Through the use of interaction terms, the effect the regression coefficient of 7 0.11 corresponded to a 0.11 of supplementation was found to be 0.29 (95% CI 0.054, lower serum retinol in boys than in girls at the mean age of 0.53; P ¼ 0.02) in children with hepatomegaly, and 0.08 13 y. Although there was no main effect of treatment, there (95% CI 0.02, 0.14; P ¼ 0.02) in children without. There was an interaction between treatment and S. mansoni were no interactions between supplementation and any of (P ¼ 0.01), but not between treatment and any of the three the four helminth infections neither in children allocated to geohelminth infections (P > 0.30). As seen, treatment was a placebo not active treatment (P > 0.10). For example, in significant predictor of serum retinol in S. mansoni-infected placebo-treated children, although the interaction term children (0.09, 95% CI 0.02, 0.16, P ¼ 0.009), but not in between supplementation and hookworm was not signifi- uninfected children (P ¼ 0.18). The regression coefficient of cant (P ¼ 0.12), the effect of supplementation was estimated S. mansoni infection of 7 0.17 corresponds to a 0.17 mmol=l separately for those with and without hookworm infection. lower serum retinol in S. mansoni-infected compared to The effect of supplementation was 0.15 (95% CI 0.06, 0.24; uninfected children at the intercept, ie among placebo P ¼ 0.002) among those with and 0.10 (95% CI: 7 0.001, treated children. 0.19; P ¼ 0.052) among those without hookworm infection. To further explore the effect of treatment of individual Neither hepatomegaly, splenomegaly, baseline malaria para- infections on serum retinol, and to express such effects in sitaemia, sickle cell phenotype nor elevated temperature terms of reductions in intensity, reductions in egg output were found to be significant predictors, or to confound the were computed for each helminth and entered into the estimated relationship between any of the predictors in the model. Only S. mansoni egg output reduction was retained final model and serum retinol at follow-up. in the model (not shown). The regression coefficient of 8Â1075 (95% CI 2Â1077,2Â1074; P ¼ 0.049) means that for each 100 epg reduction in intensity serum retinol will Discussion increase with 0.008 mmol=l. Of the 315 children allocated to Multi-micronutrient supplementation, but not multi-hel- treatment only 225 (71.4%) had S. mansoni and were treated minth chemotherapy, had a significant effect on serum with praziquantel, whereas 235 (74.6%) had at least one of retinol. Those allocated to multi-micronutrient supplemen- the three geohelminth infections and were treated with tation had a mean increase in serum retinol of 0.08 mmol=l albendazole. Albendazole alone was given to 60 (19.0%) over and above the increase seen in children allocated to children, and 50 (15.9%) received only praziquentel, while placebo supplementation. The effect of supplementation is 175 (55.6%) received both. The mean reductions in S. man- probably underestimated, since spill-over of micronutrients soni egg output in children allocated to active vs placebo to the children allocated to placebo occasionally took place. treatment were 92 and 7 128 epg, respectively. Similarly, the However, as discussed below, most of the increase unex- mean reductions in children allocated to active vs placebo plained by the interventions is thought to be due to seasonal treatment were 258 vs 7 48 epg for A. lumbricoides,93vs changes in vitamin A intake. There was no interaction 7 21 epg for hookworm and 22 vs 7 6 epg for T. trichiura. between the two interventions, ie the effect of micronutrient Interestingly, an interaction between baseline malaria supplementation did not depend on whether children were parasite intensity and sickle cell phenotype was found allocated to active or placebo treatment. (P ¼ 0.04), although neither of these variables were indepen- The effect of micronutrient supplementation was con- dent predictors. Among children with sickle cell firmed in multivariable analysis, where serum retinol was trait=anaemia, intensity of malaria parasitaemia also shown to increase with age in girls, but not in boys. A

(log10(x þ 1)) had a marginal significant negative effect on sex difference in vitamin A metabolism cannot be precluded, serum retinol ( 7 0.04, 95% CI 7 0.08, 7 0.0001; P ¼ 0.048), since serum retinol has been shown to be lower in cord blood whereas there was no effect in children without sickle cell of male compared to female neonates (Yassai & Malek, 1989), trait (0.01, 95% CI 7 0.015, 0.038, P ¼ 0.40). A marginally and even in women carrying a male compared to female significant interaction was revealed between supplementa- foetus (Friis et al, 2001). Nevertheless, since the effect of tion and malaria (P ¼ 0.06). When the analysis was repeated supplementation was similar in boys and girls, this sex after stratification for treatment, the marginal significant difference is likely to be due to differences in intake of interaction was found to be due to an interaction in children vitamin A-rich foods. Girls often take part in cooking, and allocated to placebo treatment (P ¼ 0.01) where supplemen- this may give them opportunities to get a major share of the tation was a strong predictor in children with (0.19, 95% CI vitamin A-rich foods available in the household. In contrast, 0.10, 0.28, P ¼ 0.0001), but not without ( 7 0.02, 95% CI the contribution of fishing and foraging in the bush, a

European Journal of Clinical Nutrition Micronutrients, anthelminthics and serum retinol D Mwaniki et al 671 common practice among boys, is apparently negligible. vitamin A deficiency leads to reduced humoral immune Although neither malaria parasitaemia nor sickle cell phe- response to schistosome antigens (Parent et al, 1984), and notype SA or SS were found to be predictors of serum retinol, increased worm burdens (Krakower et al, 1940) support this serum retinol decreased with increase in malaria parasitae- explanation. This discrepancy could also be due to S. mansoni mia but only in those with sickle cell trait or anaemia. infection giving rise to an acute phase response that was not Apparently, this observation has not been described before, adequately controlled for. During an acute phase response, and its significance is not clear. there is a transient decline in serum retinol because the synthesis of retinol binding protein is reduced (Rosales et al, 1996) and retinol may flux to the extravascular space S. mansoni and geohelminth infections (Thurnham & Singkamani, 1991). The decline is only par- The lack of effect of multi-helminth treatment on serum tially transient, due to increased utilisation in the tissues retinol was intriguing, especially in view of previous reports (Thurnham & Singkamani, 1991) and renal loss of retinol that S. mansoni (Friis et al, 1996,1997; Mikhail & Mansour, (Mitra et al, 1998; Stephensen et al, 1994). For example, we 1982; Sturchler et al, 1983) and A. lumbricoides (Mahalanabis have recently found that S. japonicum infection reduced et al, 1976; Sivakumar & Reddy, 1975; Tripathy et al, 1972) serum retinol but not liver retinol in the pig (Kaestel et al, infection impair serum retinol. However, through interac- 1999), considered to be a good model of not only S. japoni- tion analyses an effect of treatment was revealed in children cum (Willingham & Hurst, 1996), and maybe also of S. found to be S. mansoni infected at baseline, but not in mansoni in man. Furthermore, the extent to which estab- children found infected with any of the geohelminth infec- lished infections, or trickle- and reinfections, can elicit an tions. The effect of S. mansoni treatment corresponded to a acute phase response is not known, and whether it can 0.09 mmol=l increase in serum retinol, or almost 0.01 mmol=l explain the discrepancy in estimates of relationship between for each 100 epg reduction in egg output. This is not likely to S. mansoni and serum retinol will depend on the timing of be merely the result of a confounding effect of improved transmission in relation to blood sampling. socio-economic or health conditions, since the treatment In accordance with the results of our baseline data (Friis with praziquantel was effective in reducing S. mansoni egg et al, 1997), no effects of reduction in neither A. lumbricoides, output whereas placebo treated children had an increase in T. trichiura nor hookworm egg output on serum retinol were egg output. seen. This is in contrast to the prevailing opinion that at least Several cross-sectional studies have reported inverse asso- A. lumbricoides infection is an important cause of vitamin A ciations between S. mansoni and serum retinol (Friis et al, deficiency, although this evidence seems vague: several 1996,1997; Mikhail & Mansour, 1982; Sturchler et al, 1983). studies have demonstrated impaired vitamin A absorption Despite the the inability of drawing inferences on the pre- in children and adults (Mahalanabis et al, 1976; Sivakumar & sence and direction of a cause – effect relationship from Reddy, 1975; Tripathy et al, 1972) with A. lumbricoides infec- cross-sectional studies, some studies have interpreted the tion. However, while some studies have shown that the association as being due to S. mansoni impairing vitamin A vitamin A malabsorption is reversible upon treatment, they status (Friis et al, 1996; Mikhail & Mansour, 1982; Sturchler could not demonstrate a correlation between intensity of et al, 1983) without attempting to control for confounding infection and degree of malabsorption (Mahalanabis et al, factors, such as age, sex and the acute phase response 1976; Sivakumar & Reddy, 1975). Other studies have failed to (Mikhail & Mansour, 1982; Sturchler et al, 1983). We have detect any effects (Ahmed et al, 1993). Associations between previously reported linear negative relationship between S. A. lumbricoides infection and low serum retinol have been mansoni egg output and serum retinol from a cross-sectional found among preschool children in Panama (Taren et al, study in Zimbabwe (0.03 mmol=l reduction per 100 epg; Friis 1987) and Nepal (Curtale et al, 1994), as well as among et al, 1996), and from the baseline data of the present study Indonesian preschool children where the low serum retinol (0.07 mmol=l reduction per log epg; Friis et al, 1997), while was reversible after deworming (Jalal et al, 1998). In contrast, controlling for potential confounding factors using multi- another study among young children in Indonesia found variate regression analysis. neither effect of deworming on serum retinol nor modified In the present study, we have shown that an increase in relative dose response test (Tanumihardjo et al, 1996). None- serum retinol can be achieved through measures leading to a theless, although the evidence from previous studies is reduction in intensity of S. mansoni infection. However, the inconsistent, it should be noted that both prevalence and estimated effect is lower than what would be expected from intensities of A. lumbricoides infection were low in our study. the estimates based on the cross-sectional data from Zim- babwe (Friis et al, 1996), and the present study (Friis et al, 1997). Although the discrepancy between the estimates Seasonality could be due to chance, increased susceptibility to S. mansoni A considerable increase in serum retinol between February infection in children with vitamin A deficiency could also and October was observed, as even children allocated to contribute to the association found in the cross-sectional placebo supplementation and placebo treatment had a studies. Studies in the S. mansoni=rat model showing that 0.20 mmol=l increase in serum retinol. As previously reported,

European Journal of Clinical Nutrition Micronutrients, anthelminthics and serum retinol D Mwaniki et al 672 baseline serum retinol increased 0.05 mmol=l for each 1 y thus the worm burden of children and adults not attending increase in age (Friis et al, 1997). Thus, the 8 months increase school (Olsen, 1998). in age during the study could explain almost 0.04 of the increase. Additionally, occasional mixing of tablets between the groups may have contributed, so that the effect of Conclusion supplementation is probably underestimated. However, sea- Using schools as a system for delivering broad nutritional sonal differences in intake of vitamin A-rich foods probably and helminth control interventions was found to be feasible. accounts for most of the increase in the placebo group, and is Supplementation with multi-micronutrients was found to be in accordance with findings from other studies (Friis et al, effective in increasing serum retinol in school children. 2001; McLaren, 1993). Differences in handling, storage and Although there was no over-all effect of multi-helminth analyses of samples between the two examinations could be treatment, there was evidence that reduction in S. mansoni ruled out. Besides, similar seasonality was observed for hae- intensity of infection was able to increase serum retinol. The moglobin (unpublished data), and for haemoglobin and effect and cost-effectiveness of the interventions can be serum ferritin from a study conducted in a neighbouring further improved if higher doses of vitamin A are given. district (A Olsen, personal communication). Similarly, the effect of multi-helminth chemotherapy is expected-to be larger in populations with higher intensities of S. mansoni infection. Although no effect of treating geo- School-based health programmes helminth infections was found, it is important to note that School-based interventions are among the most cost-effec- the interventions may have effects on other important out- tive public health interventions (World Bank, 1993), since comes, such as other micronutrient deficiencies, morbidity, schools offer a pre-existing and comprehensive system for school achievement and physical activity and fitness. health delivery (Bundy & Guyatt, 1996). Additionally, school children can be approached with health education and other health-promotive activities (gardening, sanitation). Acknowledgements Due to the continuous growth and pubertal growth spurt, The study was financed by the Danish International Devel- school children have high nutritional requirements that are opment Assistance. Technical and logistical support was rarely met on a typical cereal-based diet with few animal provided by Ministry of Health, Kenya Medical Research products. Thus, school children often have deficiencies of Institute (KEMRI) and the participating schools. We grate- several micronutrients. Although micronutrient deficiencies fully acknowledge support from local education authorities may be more prevalent in preschoolers and reproductive and in particular Education Officer, Mrs C Otieno. We are women, the potential effects on school attendance and grateful to Dr W Watkins, Director of Wellcome Trust Labora- performance and, hence, scholastic achievement make tories in Nairobi for facilitating training our technicians and school children an important target group for nutritional in the set up for vitamin A assays. Also thanks to Director of interventions. Although long-term strategies should be pur- the Centre for Microbiology Research of KEMRI for availing a sued, micronutrient supplementation may be a cost-effective HPLC for use in the assays. We are indebted to the children, short-to-medium term intervention. Daily supplements their teachers and the local community for their participa- given as a tablet, or as a powder to dissolve in water as tion and cooperation. We also gratefully acknowledge the recently suggested (Ash et al, 1997), may replete the children, commitment of our technical staff and field workers. Finally, at an estimated cost of US$1 – 2 per school year. However, we would like to thank the Kenyan and the Danish KEDAHR given the modest effect of 1000 mg vitamin A, the dose Director for facilitating the research activities. This work is should probably be increased above the recommended diet- published with permission from the Director of Medical ary allowance, as the children may have higher requirements Services, Kenya, and the Director of KEMRI. due to higher infectious disease burden. This is particularly true for minerals like iron and zinc, as the bioavailability is low with a cereal-based diet. References School-based interventions to control helminth infec- Ahmed F, Mohiduzzaman M & Jackson AA (1993): Vitamin A absorp- tions may be particularly cost-effective, because the preva- tion in children with ascariasis. Br. J. Nutr. 69, 817 – 825. lence and intensity often peak at school age, and because Ash D, Tatala SR, Latham MC & Ndossi GD (1997): A micronutrient schools and surroundings are often important transmission fortified beverage supplement to combat vitamin A deficiency. Interna- sites of these excreta-related helminths. Again, long-term tional Vitamin A Consultative Group Meeting, Cairo. Bundy DAP & Guyatt H (1996): Schools for health: focus on health, strategies such as health education on sanitation and perso- education and the school-age child. Parasitol. Today 12,1– 16. nal hygiene should be central, but periodic multi-helminth Curtale F, Vaidya Y, Muhilal & Tilden RL (1994): Ascariasis, hook- chemotherapy with albendazole (effective against Ascaris, worm infection and serum retinol amongst children in Nepal. Panminerva. Med. Trichuris and hookworm) and praziquantel (effective against 36,19– 21. Fawzi WW, Chalmers TC, Herrera MG & Mosteller F (1993): Vitamin S. haematobium and S. mansoni) may be cost-effective inter- A supplementation and child mortality. A meta-analysis. J.A.M.A. ventions that may even serve to reduce transmission and 269,898– 903.

European Journal of Clinical Nutrition Micronutrients, anthelminthics and serum retinol D Mwaniki et al 673 Friis H, Mwaniki D, Omondi B, Muniu E, Magnussen P, Geissler W, Semba RD (1994): Vitamin A, immunity, and infection. Clin. Infect. Thiong’o F & Michaelsen KF (1997): Serum retinol concentrations Dis. 19, 489 – 499. and Schistosoma mansoni, intestinal helminths, and malarial para- Sivakumar B & Reddy V (1975): Absorption of vitamin A in children sitemia: a cross-sectional study in Kenyan preschool and primary with ascariasis. J. Trop. Med. Hyg. 78, 114 – 145. school children. Am. J. Clin. Nutr. 66, 665 – 671. Sommer A, Katz J & Tarwotjo I (1984): Increased risk of respiratory Friis H, Ndhlovu P, Kaondera K, Sandstrom B, Michaelsen KF, Ven- disease and diarrhea in children with preexisting mild vitamin A nervald BJ & Christensen NO (1996): Serum concentration of deficiency. Am. J. Clin. Nutr. 40, 1090 – 1095. micronutrients in relation to schistosomiasis and indicators of Stephensen CB, Alvarez JO, Kohatsu J, Hardmeier R, Kennedy JI Jr & infection: a cross-sectional study among rural Zimbabwean Gammon RB Jr (1994): Vitamin A is excreted in the urine during schoolchildren. Eur. J. Clin. Nutr. 50, 386 – 391. acute infection. Am. J. Clin. Nutr. 60, 388 – 392. Friis H, Gomo E, Kæstel P, Ndhlovu P, Nyazema N, Krarup H & Sturchler D, Holzer B, Hanck A & Degremont A (1983): The influence Michaelsen KF (2001): HIV and other predictors of serum b- of schistosomiasis on the serum concentrations of retinol and carotene and retinol in pregnancy: a cross-sectional study in retinol binding protein of a rural population in Liberia. Acta Trop. Zimbabwe. Am. J. Clin. Nutr. 73, 1058 – 1065. 40, 261 – 269. Jalal F, Nesheim MC, Agus Z, Sanjur D & Habicht JP (1998): Serum Tanumihardjo SA, Permaesih D, Muherdiyantiningsih, Rustan E, retinol concentrations in children are affected by food sources of Rusmil K, Fatah AC, Wilbur S, Muhilal, Karyadi D & Olson JA beta-carotene, fat intake, and anthelmintic drug treatment. Am. J. (1996): Vitamin A status of Indonesian children infected with Clin. Nutr. 68, 623 – 629. Ascaris lumbricoides after dosing with vitamin A supplements and Kaestel P, Lewis FJ, Willingham AL, Bogh HO, Eriksen L, Michaelsen albendazole. J. Nutr. 126, 451 – 457. KF, Sandstrom B, Hoy CE & Friis H (1999): Schistosoma japonicum Taren DL, Nesheim MC, Crompton DW, Holland CV, Barbeau I, infection and serum and tissue concentrations of retinol and zinc Rivera G, Sanjur D, Tiffany J & Tucker K (1987): Contributions in pigs. Ann. Trop. Med. Parasitol. 93, 489 – 499. of ascariasis to poor nutritional status in children from Chiriqui Krakower C, Hoffman WA & Axtmayer JH (1940): The fate of Province, Republic of Panama. Parasitology 95, 603 – 613. schistosomes (S. mansoni) in experimental infections of normal Thurnham DI & Singkamani R (1991): The acute phase response and and vitamin A deficient white rats. Puerto Rico J. Publ. Health Trop. vitamin A status in malaria. Trans. R. Soc. Trop. Med. Hyg. 85, 194 – Med. 16, 269 – 345. 199. Mahalanabis D, Jalan KN, Maitra TK & Agarwal SK (1976): Vitamin A Tripathy K, Duque E, Bolanos O, Lotero H & Mayoral LG (1972): absorption in ascariasis. Am. J. Clin. Nutr. 29, 1372 – 1375. Malabsorption syndrome in ascariasis. Am. J. Clin. Nutr. 25, 1276 – McLaren DS (1993): Vitamin A. In Human Nutrition and Dietetics,eds. 1281. JS Garrow and WPT James, Vol 9,pp208– 217. Edinburgh: West KP Jr, Katz J, Khatry SK, LeClerq SC, Pradhan EK, Shrestha SR, Churchill Livingstone. Connor PB, Dali SM, Christian P, Pokhrel RP & Sommer A (1999): Mikhail MM & Mansour MM (1982): Complications of human Double blind, cluster randomised trial of low dose supplementa- schistosomiasis and their effect on levels of plasma copper, zinc tion with vitamin A or beta carotene on mortality related to and serum vitamin A. Hum. Nutr. Clin. Nutr. 36, 289 – 296. pregnancy in Nepal. The NNIPS-2 Study Group. Br. Med. J. 318, Mitra AK, Alvarez JO, Guay-Woodford L, Fuchs GJ, Wahed MA & 570 – 575. Stephensen CB (1998): Urinary retinol excretion and kidney Willingham AL & Hurst M (1996): The pig as a unique host model for function in children with shigellosis. Am. J. Clin. Nutr. 68, Schistosoma japonicum infection. Parasitol. Today 12, 132 – 134. 1095 – 1103. World Bank (1993): World Development Report: Investing in Health. Olsen A (1998): The proportion of helminth infections in a commu- Oxford: Oxford University Press. nity in western Kenya which would be treated by mass chemother- World Health Organization (1991): Basic Laboratory Methods in Med- apy of schoolchildren. Trans. R. Soc. Trop. Med. Hyg. 92, 144 – 148. ical Parasitology. Geneva: WHO. Parent G, Rousseaux-Prevost R, Carlier Y & Capron A (1984): Influ- Yassai MB & Malek F (1989): Newborns vitamin A in relation to sex ence of vitamin A on the immune response of Schistosoma man- and birth weight. J. Trop. Pediatr. 35, 247 – 249. soni-infected rats. Trans. R. Soc. Trop. Med. Hyg. 78, 380 – 383. Rosales FJ, Ritter SJ, Zolfaghari R, Smith JE & Ross AC (1996): Effects of acute inflammation on plasma retinol, retinol-binding protein, and its mRNA in the liver and kidneys of vitamin A-sufficient rats. J. Lipid Res. 37, 962 – 971.

European Journal of Clinical Nutrition