Professional Documents
Culture Documents
& 2004 Nature Publishing Group All rights reserved 0954-3007/04 $30.00
www.nature.com/ejcn
ORIGINAL COMMUNICATION
Effectiveness of weekly vitamin A (10 000 IU) and iron
(60 mg) supplementation for adolescent boys and
girls through schools in rural and urban East Java,
Indonesia
DD Soekarjo1,2*, S de Pee3, JA Kusin2, WHP Schreurs2, W Schultink4, Muhilal5 and MW Bloem3
1
Helen Keller International Indonesia, Jakarta, Indonesia; 2University of Amsterdam, Netherlands; 3Helen Keller International,
Singapore; and 4UNICEF, New Delhi, India; and 5Nutrition Research and Development Centre, Bogor, Indonesia
Objective: High prevalences of vitamin A deficiency and anaemia among adolescents warrant interventions. This study
evaluated the effectiveness of school-based supplementation to reduce anaemia and improve vitamin A status.
Design: School-based, grade-randomized, intervention.
Subjects and setting: In all, 1757 girls and 1859 boys, aged 12–15 y, in 24 Junior High Schools.
Interventions: Weekly supplementation for 14 weeks with 60 mg iron and 250 mg folate (Fe group; n ¼ 978), 10 000 IU vitamin
A (VA group; n ¼ 970) or both (VAFe group; n ¼ 1042) to subjects in 15 schools, compared to subjects in nine other schools not
receiving supplements (control; n ¼ 626).
Results: The baseline anaemia prevalence (Hb o120 g/l) in girls was 20% (prepubertal) and 26% (pubertal), and in boys 24%
(pre-pubertal) and 11% (pubertal). Serum retinol concentrations were low (o1.05 mmol/l) in 41% of boys and 45% of girls. The
interventions did not increase haemoglobin concentrations. Serum retinol concentration of boys, but not girls, in the VA group
increased (0.33 vs 0.07 mmol/l in controls; Po0.01). The risk factors for low serum retinol concentration were lower baseline
serum retinol concentration (OR 0.02–0.03) with, for girls, nightblindness at baseline (OR 5.88), and for boys, not receiving
vitamin A (OR control: 1.00; VA: 0.37; Fe: 0.77; VAFe: 0.34) and maternal illiteracy (OR mother never attended school 1.00,
mother received any formal education 0.17–0.33).
Conclusions: Supplementation with vitamin A increased serum retinol concentration of boys. Iron supplementation did not
change Hb. This appeared to be due to poor compliance, and partly related to side effects.
Sponsorship: This study was funded by USAID through the OMNI Project.
European Journal of Clinical Nutrition (2004) 58, 927–937. doi:10.1038/sj.ejcn.1601914
Introduction
Anaemia and vitamin A deficiency are very prevalent
*Correspondence: DD Soekarjo, Helen Keller International, PO Box 4338, throughout the world. Globally, 55% of underfives and
Jakarta Pusat, Indonesia.
56% of pregnant women are anaemic (ACC/SCN, 1997). In
E-mail: ddsoekarjo@hkiasiapacific.org
Guarantor: MW Bloem. Indonesia, prevalences are similar (UNICEF, 1996). Among
Contributors: DDS supervised and conducted the field work, Asian adolescents, anaemia prevalences have been found to
performed preliminary statistical analyses and drafted the paper. be range from 11 to 26% (Sunethra Atukorala & De Silva,
SdP conducted further statistical analyses and contributed to writing
1990; Ahmed et al, 1996; Angeles-Agdeppa et al, 1997;
the paper. JAK and WHPS gave critical comments on the interpreta-
tion of the results and contributed to the writing of the paper. Soekarjo et al, 2001). Vitamin A deficiency is also prevalent
WS contributed to the design and gave comments on the draft in Indonesian preschool children (51% serum retinol
versions of the manuscript. M gave critical comments on the design of o0.70 mmol/l; Muhilal & Permaesih, 1994), and pregnant
the study and interpretation of the results. MWB contributed
(33.5%, Suharno et al, 1992), lactating (36%; de Pee et al,
substantially to the design and analytic strategy.
Received 17 January 2003; revised 21 September 2003; accepted 3 1995) and adolescent women (30% o0.70 mmol/l; Angeles-
October 2003 Agdeppa et al, 1997).
Vitamin A and iron supplements for adolescents
DD Soekarjo et al
928
The main cause of anaemia and vitamin A deficiency is Menengah Pertama) and seven Islamic Junior High Schools (U-
inadequate intake of bioavailable iron and vitamin A, in MTs ¼ urban Madrasah Tsanawiyah). Pupils in all selected
addition to other factors like excessive blood loss, frequent schools were of middle and low socioeconomic class, of
infections and other micronutrient deficiencies. Both con- Javanese or Madurese ethnicity, and Muslim, except for one
ditions have negative consequences for health, develop- U-SMP, where the majority was Christian Chinese.
ment, well-being and productivity (Delisle et al, accessed Except in the control schools, all pupils received supple-
2002). In addition, vitamin A deficiency plays a role in the ments. However, data were collected only on pupils who
development of anaemia (Mejia et al, 1977; Bloem, 1995), were between 12 and 15 y old at the start of the study, and
and vitamin A supplementation can enhance the effect of whose parents gave written informed consent for taking
iron supplementation (Mejia & Chew, 1988; Suharno et al, supplements and being enrolled in the data collection
1993). (91.4% of eligible pupils; n ¼ 5166).
Adolescence (the period between the ages of 10 and 19 y, This study was conducted between October 1996 and May
according to the WHO definition) is the period of the third- 1997, in collaboration with the Indonesian Ministries of
fastest growth a human being experiences, after the foetal Health and Education after approval from the Medical
and infancy periods (Tanner, 1978). This rapid growth highly Ethical Committee of the Ministry of Health, Indonesia.
increases the physiological needs for micronutrients such as
iron and vitamin A (Brabin & Brabin, 1992; Antilla & Siimes,
1996), and thus the susceptibility to deficiency in these Sample size
phases of the lifecycle. As important changes (physiological, Based on the sparse data available at the time on anaemia
psychological and social) occur, and many adult patterns are prevalence among Indonesian adolescent girls (Schultink,
established, adolescence provides a window of opportunity 1995, personal communication), an anaemia prevalence of
to prevent adult health problems. However, adolescents’ 30% was presumed. In order to be able to detect a 50%
needs are not met by the health services, especially in reduction from this prevalence at a ¼ 0.05 and b ¼ 0.80, and
developing countries. accounting for an expected dropout rate of 25%, 168 subjects
Iron supplementation programmes focus mainly on per intervention group were needed (total 672).
pregnant women, while vitamin A capsule distribution
programmes for underfives and postpartum women exist in
Indonesia as in many other countries. Supplementation Design
during adolescence might be more effective as it could take Adolescents from 15 schools (four U-MTs, seven U-SMP and
place over a relatively long period of time, and weekly dosing four R-SMP) (n ¼ 2990) were randomly selected to receive
would be sufficient (Fogelholm et al, 1994; Viteri, 1996; weekly supplements, while adolescents in the other nine
Angeles-Agdeppa et al, 1997). Schools and worksites may be schools (three U-MTs, two U-SMP and four R-SMP) served as
appropriate distribution channels, while teachers and peer controls (n ¼ 1750). In each of the schools receiving supple-
group support to promote behavioural change might lead to ments, each of the three grades was randomly allocated to
higher compliance than found among pregnant women. receive one of the three supplementation regimes: weekly
The present study was aimed to test the feasibility and 10 000 IU vitamin A (VA group; n ¼ 970), weekly 60 mg
effectiveness of weekly vitamin A and iron supplementation elemental iron (as ferrous sulphate) plus 250 mg folate (Fe
when delivered in a programmatic context to adolescent group; n ¼ 978) or weekly 10 000 IU vitamin A and 60 mg
girls and boys attending Junior High Schools in rural and elemental iron plus 250 mg folate (VA þ Fe groups; n ¼ 1042)
urban East Java, Indonesia. It was implemented by Helen (Figure 1). Owing to the theoretical possibility of pregnancy
Keller International in cooperation with the Ministry of occurring in the study population, the supplements used in
Health of the Government of Indonesia as part of the GIRLS the intervention complied with the official recommenda-
(Gizi: Intervensi Remaja Lewat Sekolah/Nutrition Interven- tions of the Indonesian government for safe use of supple-
tion for Adolescents through Schools) Project. ments in pregnancy. Therefore, the dosage of vitamin A did
not exceed 10 000 IU and iron-folate tablets were used, such
as are routinely distributed to pregnant women in Indonesia.
Methods The red iron/folic acid tablets and yellow sugar-coated
Subjects vitamin A tablets were all produced locally (Kimia Farma,
24 schools were selected to represent the two main school Jakarta, Indonesia). All pupils were aware of which supple-
types in Indonesia as well as both rural and urban locations. ment(s) they were taking and they were told that the
Eight rural general Junior High Schools (R-SMP ¼ rural supplements would improve their health and prevent/cure
Sekolah Menengah Pertama) were randomly selected from anaemia. This information was also disseminated to the
two districts (Bangkalan and Sampang) on the island of parents by means of a leaflet, which was part of the informed
Madura. The urban area was represented by 16 randomly consent procedure.
selected schools in Surabaya, the capital of East Java: nine Field workers supplied drinking water and the pupils took
urban general Junior High Schools (U-SMP ¼ urban Sekolah the supplements in their classrooms. One or two field
Data
presented Control (n=626) VA (n=970) Fe (n=978) VAFe (n=1042)
boys: 1859 Boys n=317 Boys n=507 Boys n=490 Boys n=545
girls: 1757 Girls n=309 Girls n=463 Girls n=488 Girls n=497
Total: 3626
Subsample
for serum Control (n=128) VA (n=175) Fe (n=154) VAFe (n=168)
retinol Boys n=74 Boys n=85 Boys n=68 Boys n=92
boys: 319 Girls n=54 Girls n=90 Girls n=86 Girls n=76
girls: 306
Total: 625
Figure 1 Supplementation study design (VA ¼ vitamin A; Fe ¼ iron; VA þ Fe ¼ vitamin A and iron; SMP ¼ general junior high school;
MTs ¼ religious junior high school).
percentage of pubertal subjects, vitamin A intake from among boys in the iron group compared to the VA group,
animal sources and the level of maternal education. Boys and the mean serum retinol concentration was higher in
in the four groups did not differ with regard to the boys in the VA þ Fe groups than in the VA group.
percentage of pubertal subjects, vitamin A intake from Differentiation between pubertal and prepubertal subjects
animal and plant sources and anthropometric indicators. is not shown in Table 1a and b, as this would make the table
There was no difference in vitamin A intake from either very complicated, but the proportions of pubertal subjects in
animal or plant sources between girls and boys. each group are indicated and were similar in the four groups
Paternal educational level and anthropometric indicators for both boys and girls. Anaemia prevalence was 20% among
were slightly worse for girls in the control and VA þ Fe groups prepubertal and 26% among pubertal girls, 24% among
compared to the other two groups; however, this was not prepubertal and 11% among pubertal boys. There was a
related to differences in any of the micronutrient-related positive relationship between pubertal status and haemoglo-
indicators (haemoglobin and serum retinol concentrations, bin concentration in boys (129713 g/l in prepubertal boys vs
and nightblindness). Boys in the control group had parents 136714 g/l in pubertal boys, Po0.001), but not in girls
with lower education than in the other groups, indicative of (128711 vs 126712 g/l).
lower socioeconomic status. Serum retinol concentrations were o1.05 mmol/l in 41.4%
There were no differences in any of the indicators of of boys and 45.1% of girls. The proportions of subjects with
micronutrient status among the girls, while among boys serum retinol concentrations o0.70 mmol/l were 11.4% boys
there were no differences among the groups in the (range 6.5–10.8%) and 7.7% girls (range 4.4–9.3%), and did
prevalence of anaemia or nightblindness. However, the not differ between the four groups in either sex. There was a
mean haemoglobin concentrations were slightly higher positive relationship in girls between puberty and serum
Control (n ¼ 309) Vitamin A (n ¼ 463) Iron (n ¼ 488) Vitamin A and iron (n ¼ 497) P-value
(a) Girls
Age (y)a 14.4 (13.6; 15.0)B 14.3 (13.5; 14.9)B 14.3 (13.5; 15.0)B 13.9 (13.2; 14.7)A o0.001
% that had reached pubertyb 82.5 (255) 83.6 (387) 85.0 (415) 81.1 (403) NS
Weight (kg)c 38.976.7A 41.478.6B 40.978.2B 39.576.9A o0.001
Height (cm)c 147.676.2A 148.976.1B 148.875.7B 148.775.5B o0.05
BMI (kg/m2)c 17.872.4A 18.673.1B 18.473.1B 17.872.6A o0.001
VA intake from retinol sources (RE)a 175 (60;350) 175 (60;375) 185 (60; 360) 180 (60; 350) NS
VA intake from plant foods (RE)a 50 (20; 225) 100 (20; 375) 80 (20; 376) 90 (30; 295) o0.05
Fathers with primary education or lessb,d 58.0B 47.6A 47.1A 47.9B o0.05
Mothers with primary education or lessb,e 76.2 65.7 66.0 65.6 NS
Control (n ¼ 317) Vitamin A (n ¼ 507) Iron (n ¼ 490) Vitamin A and iron (n ¼ 545) P-value
(b) Boys
Age (y)a 14.2 (13.6; 14.9)BC 14.1 (13.4; 14.7)B 14.3 (13.6;15.1)C 14.0 (13.3; 14.7)A o0.001b
% that had reached pubertyb 67.2 (213) 70.0 (355) 72.2 (354) 68.8 (375) NS
Weight (kg)c 40.678.3 40.8710.4 41.779.4 41.179.6 NS
Height (cm)c 152.078.6 152.179.8 153.379.4 152.478.8 NS
BMI (kg/m2)c 17.472.5 17.472.8 17.572.5 17.572.8 NS
VA intake from retinol sources (RE)a 175 (60;350) 185 (70; 350) 185 (60;375) 180 (60;360) NS
VA intake from plant foods (RE)a 80 (20;295) 90 (20;385) 80 (20;385) 90 (20;370) NS
Fathers with primary education or lessb,d 68.2B 53.1A 53.1A 49.0A o0.001
Mothers with primary education or lessb,e 77.8c 66.5AB 70.5BC 62.5A o0.001
retinol concentration: 1.0670.32 vs 1.1970.43 mmol/l in the groups that received only one type of supplement. More
prepubertal and pubertal girls (Po0.05). The same trend was girls than boys reported to have taken at least 50% of the
seen in boys, although it was not significant: 1.1170.43 vs iron tablets without supervision (70.1 vs 50.7%; Po0.05),
1.2170.45 mmol/l in prepubertal and pubertal boys, respec- but there was no such difference for vitamin A (70.1 vs
tively (NS). 61.2%; NS).
After the supplementation was over, self-reported compli- There was a nonsignificant relationship between reported
ance for iron supplements was collected from 136 boys and side effects (in particular, nausea, vomiting and bad taste/
134 girls and for vitamin A supplements from 134 boys and smell often associated with nausea, which together are
137 girls. Almost all respondents (90–97%) claimed to have referred to here as gastrointestinal (GI) side effects) and
taken at least half the tablets distributed under supervision reported compliance among girls. Girls who reported GI side
(data not shown). Without supervision, however, this effects of iron tablets (41.4% of 134 interviewed girls)
proportion dropped to 50–70% (Figure 2). Self-reported reported a lower compliance to iron tablets when there was
compliance without supervision was higher among girls no supervision than girls without GI side effects: 60.0 vs
than among boys, and higher in the VA þ Fe groups than in 76.9% reported having taken at least 50% of the tablets. No
Girls
Prepubertal (297)
Nonanaemic (237) 2.7 (6.2; 0.8) (46) 4.9 (7.4; 2.4) (58) 5.6 (8.0; 3.2) (58) 5.0 (7.6; 2.4) (75)
Anaemic (60) 9.0 (4.4; 13.6) (8) 5.9 (0.9; 11.0) (18) 7.5 (0.5; 14.6) (15) 10.2 (6.1; 14.3) (19)
Pubertal (1460)
Nonanaemic (1088) 6.1 (7.7; 4.5) (189) 7.0 (8.4; 5.6) (284) 7.1 (8.3; 5.8) (307) 6.3 (7.6; 5.0) (308)
Anaemic (372) 5.6 (2.5; 8.6) (66)AB 2.7 (0.5; 4.9) (103)A 7.8 (5.7; 9.9) (108)B 4.4 (1.9; 6.9) (95)AB
Boys
Prepubertal (562)
Nonanaemic (426) 4.3 (7.1; 1.5) (72) 4.0 (6.1; 1.8) (119) 6.1 (8.3; 4.0) (102) 3.3 (5.3; 1.4) (133)
Anaemic (136) 7.5 (3.8; 11.1) (32) 8.4 (3.5; 13.3) (33) 5.3 (1.2; 9.4) (34) 7.1 (3.1; 11.1) (37)
Pubertal (1297)
Nonanaemic (1161) 3.1 (4.8; 1.3) (194) 3.8 (5.3; 2.3) (308) 5.5 (6.8; 4.2) (325) 3.5 (4.9; 2.1) (334)
Anaemic (136) 9.8 (5.2; 14.4) (19) 12.0 (8.6; 15.4) (47) 7.4 (3.4; 11.4) (29) 12.9 (8.5; 17.3) (41)
A,B
Within a row, significant differences exist between groups that do not share the same letter (Po0.05, ANOVA with Bonferroni correction for post hoc multiple
comparisons).
mother with no formal education, and received either no found results resembling those documented by Suharno et al
supplementation or iron only, in addition to having a lower (1993) in pregnant women. However, the subjects of both
baseline serum retinol concentration. studies were older, were all postmenarcheal, had lower serum
retinol concentrations and, in Ahmed et al’s study, were all
anaemic, which can explain part of the difference in impact.
Discussion Tee et al found an increase in haemoglobin concentration of
This school-based intervention among 12–15 y-old adoles- anaemic girls after iron/folate supplementation, but there
cents found no effect on haemoglobin concentration of 14 was no placebo group to compare this with. Among
weeks of supplementation with 60 mg iron þ 250 mg folate nonanaemic girls, weekly supplementation with 120 mg
and 10 000 IU vitamin A, either alone or in combination. but not 60 mg iron increased haemoglobin concentration.
While there was no effect of the supplementation on the Their subjects were older, and no details on pubertal or
serum retinol concentration of girls, boys showed an socioeconomic status were provided. Recent studies among
increase in serum retinol concentration after supplementa- pregnant women also did not find an additional benefit of
tion with vitamin A alone but not with iron. vitamin A supplementation on haemoglobin concentrations
At the time this study was designed, not much literature in pregnancy nor postpartum, nor of postpartum serum
was available on iron and vitamin A supplementation among retinol concentrations (Muslimatun et al, 2001a, Muslima-
adolescents. However, based on supplementation studies in tun et al, 2001b, Semba et al, 2001).
other groups like children (Angeles et al, 1993) and pregnant The interventions were randomized per grade and the
women (Suharno et al, 1993), it was deemed reasonable to proportion of pupils attending the different types of schools
expect an impact in adolescents as well. Lately, other studies was similar in all four treatment groups. The control group
have shown the efficacy of weekly supplementation with consisted of pupils from similar schools in the same area who
iron and/or vitamin A to increase haemoglobin (and serum did not receive any intervention and were measured at the
retinol) concentrations among adolescent girls (Angeles- same periods as the intervention groups. Differences
Agdeppa et al, 1997, Tee et al, 1999, Ahmed et al, 2001). No between schools could thus not influence the results of the
studies among boys were found. The differences in findings intervention. No difference was found between the control
with the present study may be explained by the following group and the intervention groups, and there is no reason to
critical review of the various factors, which can cause assume this would have been different had a placebo been
differences in impact of the supplementation on haemoglo- used. Mix up of supplements between the different groups
bin and serum retinol concentration of adolescents. was highly unlikely because the supplements had different
The main difference is that those studies were efficacy colours and no placebo was used.
trials and therefore only reported on compliers, while the The study coincided with Ramadan, the Muslim
study presented here assessed the effectiveness of supple- fasting month. This could not be avoided, because
mentation in a programmatic setting. Only two studies otherwise the time between school examinations
included vitamin A in the intervention (Angeles-Agdeppa (during which no supplementation and/or data collection
et al, 1997; Ahmed et al, 2001), and in particular the latter activities were allowed in the schools) would have been
Table 3 Changes in serum retinol concentrations (mmol/l) in girls (n ¼ 306) and boys (n ¼ 319) by initial serum retinol concentration status and initial pubertal status (mean (95% confidence
Girls Overall Overall (306) 0.14 (0.05; 0.23) (54) 0.14 (0.06; 0.22) (90) 0.09 (0.01; 0.17) (86) 0.14 (0.02; 0.26) (76)
Pre (47) 0.17 (0.11; 0.45) (7) 0.15 (0.06; 0.36) (17) 0.09 (0.23; 0.41) (8) 0.30 (0.07; 0.66) (15)
Post (259) 0.13 (0.03; 0.23) (47) 0.13 (0.05; 0.22) (73) 0.09 (0.01; 0.18) (78) 0.10 (0.02; 0.23) (61)
Z1.05 mmol/l Overall (168) 0.07 (0.17; 0.03) (22) 0.06 (0.05; 0.18) (54) 0.01 (0.10; 0.11) (45) 0.03 (0.14; 0.21) (47)
Pre (17) (0) 0.11 (0.18; 0.95) (4) 0.07 (0.51; 0.65) (5) 0.31 (0.45; 1.08) (8)
Post (151) 0.07 (0.17; 0.03) (22) 0.08 (0.03; 0.19) (50) 0.00 (0.12; 0.11) (40) 0.02 (0.18; 0.14) (39)
DD Soekarjo et al
o1.05 mmol/l Overall (138) 0.28 (0.16; 0.40) (32) 0.24 (0.14; 0.35) (36) 0.19 (0.07; 0.30) (41) 0.31 (0.18; 0.44) (29)
Pre (30) 0.17 (0.11; 0.45) (7) 0.23 (0.05; 0.41) (13) 0.13 (0.53; 0.79) (3) 0.28 (0.08; 0.48) (7)
Post (108) 0.31 (0.17; 0.45) (25) 0.25 (0.12; 0.39) (23) 0.19 (0.07; 0.31) (38) 0.32 (0.15; 0.49) (22)
Boys Overall Overall (319)c 0.07 (0.01; 0.15) (74)A 0.33 (0.22; 0.45) (85)B 0.17 (0.08; 0.27) (68)AB 0.19 (0.07; 0.30) (92)AB
Pre (83) 0.08 (0.06; 0.22) (20) 0.27 (0.07; 0.46) (26) 0.17 (0.03; 0.38) (18) 0.33 (0.12; 0.53) (19)
Post (236)d 0.07 (0.03; 0.17) (54)A 0.37 (0.22; 0.51) (59)B 0.17 (0.06; 0.28) (50)AB 0.15 (0.02; 0.29) (73)AB
Z1.05 mmol/l Overall (187) 0.02 (0.12; 0.08) (48) 0.11 (0.06; 0.28) (37) 0.19 (0.02; 0.35) (35) 0.08 (0.06; 0.22) (67)
Pre (42) 0.03 (0.26; 0.32) (9) 0.06 (0.45; 0.33) (10) 0.13 (0.22; 0.48) (11) 0.31 (0.01; 0.64) (12)
Post (145) 0.03 (0.14; 0.07) (39) 0.17 (0.02; 0.37) (27) 0.21 (0.02; 0.40) (24) 0.03 (0.13; 0.19) (55)
o1.05 mmol/l Overall (132)e 0.25 (0.14; 0.36) (26)AB 0.51 (0.37; 0.65) (48)C 0.16 (0.05; 0.26) (33)A 0.47 (0.33; 0.60) (25)BC
Pre (41)d 0.12 (0.04; 0.27) (11)A 0.47 (0.31; 0.63) (16)B 0.24 (0.07; 0.41) (7)AB 0.35 (0.10; 0.59) (7)AB
Post (91)f 0.35 (0.20; 0.49) (15)AB 0.53 (0.32; 0.74) (32)B 0.13 (0.01; 0.26) (26)A 0.52 (0.34; 0.69) (18)B
a
Baseline serum retinol concentration.
b
Pre ¼ pre-pubertal; Post ¼ pubertal.
c
Po0.01.
d
Po0.05.
e
Po0.001.
f
Po0.005.
A,B,C
Within a row, significant differences exist between groups that do not share the same letter (Po0.05, ANOVA with Bonferroni correction for post-hoc multiple comparisons).
Vitamin A and iron supplements for adolescents
DD Soekarjo et al
935
Table 4 Odds ratios (OR) (95% confidence interval (CI)) for having a serum retinol concentration o1.05 mmol/l postintervention: forward entrance into
the multiple logistic regression modelsa
Baseline serum retinol concentration (mmol/l ) 0.02 (0.01–0.07) o0.0001 0.03 (0.01–0.09) o0.0001
Nightblindness at baseline 5.88 (1.21–28.6) o0.05
Intervention o0.05
Control 1.00
Vitamin A 0.37 (0.16–0.87) o0.05
Iron 0.77 (0.32–1.85) NS
Vitamin A and iron 0.34 (0.14–0.82) o0.05
insufficient for the supplementation to be effective. During group compared to those in the VA þ Fe groups was expected
Ramadan people generally consume more meat and other because of their lower baseline serum retinol concentrations.
sources of haem iron, which might have led to higher Owing to the similar baseline concentrations of girls and
haemoglobin concentrations. However, due to the double boys, and the lower recommended daily allowance (RDA), a
randomization scheme, there is no reason to assume that similar increase would have been expected in girls. It is
this effect, if at all present, differed between intervention possible that the RDA for girls is either not correct or not
groups. applicable in this population, that subclinical infections
During the fasting month and holidays supplements had were more prevalent among girls or that there were other
to be taken at home without supervision (n ¼ 5 tablets). Self- factors selectively decreasing absorption in girls, despite the
reported compliance without supervision was low, but similar conditions of supplement consumption. However,
similar to what Angeles-Agdeppa et al (1997) found for another factor that potentially could explain the lack of
supplements taken at home. Thus, it should be assumed that impact particularly among the girls is poor compliance.
the actual duration of supplementation in this study was In view of the lack of impact found, it is highly probable
only 9 weeks, which was probably too short for the that respondents over-reported compliance rates, and that
supplements to have an effect. this occurred more among girls than among boys. This is
The fact that vitamin A supplements were not effective in confirmed by reports from field workers that supplements
increasing serum retinol concentration in girls and haemo- were found on the floors of the schools, indicating that even
globin concentration in both girls and boys might be due to under supervision, actual compliance was lower than
the fact that the dose was relatively low. A measure of reported, especially for iron. Thus, self-reporting is not a
10 000 IU on a weekly basis amounts to only 428 retinol reliable tool to measure compliance. Almost all subjects
equivalents (RE)/day, while the RDA for vitamin A in this age claimed to have taken a considerable number of the
group is 700 RE for girls aged 14–18 y and 900 RE for boys of supplements and as a result, there was no relationship
the same age (US Institute of Medicine, 2000). between self-reported compliance and impact. This phenom-
However, short duration and low vitamin A dose alone enon of over-reporting compliance has also been found
cannot completely explain these results. In boys with low among Indonesian pregnant women (Schultink et al, 1993;
serum retinol concentrations, the increase in serum retinol Ridwan et al, 1996).
concentration in the two groups receiving vitamin A was Most studies to date have focused on anaemic subjects.
almost twice as large as in the other two groups (and also However, from a public health point of view, it may be
compared to the girls). In addition, there was a nonsignifi- worthwhile to target the whole population to have
cant increase in haemoglobin concentration among pubertal an impact among a subgroup, rather than identify subjects
anaemic boys in these treatment groups, showing that the and then specifically target those. Therefore, this study
dose was probably adequate for boys. included all pupils in the schools, but analyses were
The larger increase in serum retinol concentrations of boys performed with the aim to determine differential effects in
with low initial serum retinol concentrations in the VA subgroups.