MATERIALS AND METHODS
The women included in this study were participants in a large case-control study in Brazil. Details of this population are given elsewhere (3). Blood samples and FFQs were available in 710 pregnant women between March 1991 and January 1992. Women were selected from 4 hospitals in Campinas city, São Paulo state, 12-72 hours after delivery. None of them received vitamin A supplements during pregnancy.

Womens blood sample was taken by Becton Dickinson vacutainer tubes. Vitamin A was determined in plasma, in duplicate, by high performance liquid chromatography (HPLC).

An interviewer-administered FFQ was utilised, considering the high percentage of illiteracy. The FFQ included 55 foods, and was based on the typical diet of the low socioeconomic families in the Southeast of Brazil (4). For calculation purposes, first were identified all potential sources of retinol and carotenoids; the portion sizes being assigned to each food item on the basis of a previous study in the same area (2). The FFQ included the foods that provide a minimum of 5% of the total average of retinol equivalents (RE) recommended for this population, which in quantitative terms, represents a value of 40 ug of RE for a small portion.

To classify the foods on high, moderate and low concentrations of RE we used two selection criterias: the first one was based on a previous study of vitamin A content in regional foods (5), and the second was based on the food composition table of McCance & Widdowson's (6).

Daily, weekly and monthly frequencies of consumption of foods with high, moderate and low concentrations of vitamin A were grouped according to the IVACG recommendations (1), in usual pattern of food consumption (UPF) scores, according to the average frequency they consumed each food in the 9 months of pregnancy. Women were allocated by these scores in 3 different categories (high, moderate and low) according to their risk of developing vitamin A deficiency. The categories of risk were based on the RDA safe level of vitamin A intake of 800 RE/day (5600 RE/week) (7), and the basal requirement of 370 RE/day for pregnant women (8). The cut-off points for the high, moderate and low categories of risk for vitamin A deficiency, based on scores, were ≤ 3/5 of the RDA recommendation (score = 480) in the low category, between 3/5 ofthe RDA recommendation and 3/5 of the basal requirement for vitamin A (score = 220) in the moderate category, and ≤ 3/5 of the basal requirement in the high category (1).

Univariate analyses were used in EPI-INFO to identify the women that ingested foods with high, moderate and low concentrations of vitamin A, based on the food composition table of McCance & Widdowsons (6), and to identify possible differences between the plasma levels of vitamin A and the FFQ scores as indicators of vitamin A deficiency. The correlation between the plasma levels of vitamin A and the FFQ scores were assessed in the SPSS Statistical Software, utilising the Spearman correlation coefficient.

Informed written consent was obtained from all subjects and the protocol was approved by the ethical committee of the four hospital s in Campinas.

RESULTS
According to Table 1, the mean level of vitamin A in plasma was 1.7mmol/l (SD = 0.59); the mean value for the FFQ score was 484.31 (SD = 117.88).

Six hundred and twenty eight (628) women had plasma levels of vitamin A > 1.05mmol/l, and almost half of them (315) had low risk for vitamin A deficiency, according to the score categories. Seventy three women had plasma levels of vitamin A between 0.71-1.05mmol/1 (marginal levels of vitamin A) and 43 of them had moderate risk for vitamin A deficiency, by the score categories. Only nine women had concentrations of vitamin A ≤ 0.70mmol/l (deficient levels), none of them had high risk, by the score categories. There was no statistically significant difference between the biochemical indicator of vitamin A deficiency, and the FFQ scores (p = 0.19) (Table 2).

Figure 1 shows the distribution of the plasma levels of vitamin A and the FFQ scores. There was a weak statistically significant correlation between the plasma levels of vitamin A and the FFQ score categories (r = 0.11; P = 0.005).

The most frequently ingested foods with high, moderate and low concentrations of vitamin A are presented in Figure 2. The staple diet of the Brazilian pregnant women in this region is given in Figure 3.

TABLE 1
Mena and standard deviation (SD) of plasma
levels of vitamin A (umol/l) and FFQ scores

TABLE 2
Categories of risk for vitamin A deficiency based
on plasma levels of vitamin A and FFQ scores

FIGURE 1
Distribution of plasma levels of vitamin A versus FFQ scores

FIGURE 2
Foods with high, moderate and low concentration of vitamin A,
most frequently ingested by the population

FIGURE 3
Staple diet of the Brazilian pregnat women in the region

DISCUSSION
There were very few studies in the literature comparing the intake of vitamin A, as reporte on a food frequency questionnaire with the corresponding biochemical indicator of vitamin A status. In countries where vitamin A deficiency is not a public health problem, plasma carotenoids showed a higher statistically significant correlation with dietary questionnaires (although weak), than plasma level of retinol (11-15). However, in popu1ations where vitamin A deficiency was preva1ent, plasma retinol was associated with a poor dietary intake of vitamin A (16).

Despite all the ease of the administration of the FFQ, we were aware of its limitations such as under or overestimation of the frequency of food intake (9). The diet of the pregnant women included in this study was similar to those obtained in ENDEF (4,10), a large national study that assessed the most common type of foods purchased per family in several Brazilian regions and large cities.

Our results indicate that the FFQ does not provide a very precise information on vitamin A status, probably because of the day today variation in vitamin A intake, and the insensitivity of plasma concentrations to consumption of vitamin A. However, by using a simple score based on the FFQ, we were able to predict that the majority of the mothers were not at high risk for vitamin A deficiency.

ACKNOWLEDGMENTS
This work was supported by awards from the Overseas Development Agency (ODA) of Great Britain, and from Conselho Nacional de Desenvolvimento Científico e Tecno- lógico (CNPq) of Brazi1. The authors gratefully acknowledge Dr. Fernando Pavani, Dr. Roberto Franco do Amara1, Prof. A. Barros, Dr. Dulce Telini and Dr. Sergio Bezerra from the Maternidade de Campinas, Universidade Estadual de Campinas, Pontifícia Universidade Católica de Campinas and Hospital A1bert Sabin for allowing to carry out the project in these hospitals, and to Ms Rebecca Abbott for measuring the vitamin A in plasma. We are also grateful for the excellent assistance provided by the field workers: Dr Ivana Feliz Augusto, Rosa María Vieira de Carvalho, Miriam Kubo and Sergia Haddad.

REFERENCES

1. IVACG. Guidelines for the development of a simplified dietary assessment to identify groups at risk for inadequate intake of vitamin A. Report of the International Vitamin A Consultative Group (IVACG) Washington, 1989.
   
   
2. Slater BV. Identificação do risco de deficiência de vitamina A em gestantes, usando metodologia simplificada de inquérito alimentar (Dissertação). São Paulo: University of São Paulo, 1996 (in portuguese).
   
   
   
3. Rondón PHC, Abbott R, Rodrigues L, Tomkins A. The influence of maternal nutritional factors onintrauterine growth retardation in Brazil. Paediatr Perinat Epidemiol1977;11 :152-166.
   
   
4. ENDEF (Estudo Nacional da Despesa Familiar), Consumo alimentar-Antropometria, dados preliminares, região II- São Paulo, região IV-Minas Gerais e Espíritu Santo. Secretaria de Planejamento da Presidencia da República, Fundação Instituto Brasileiro de Geografia e Estatística. Rio de Janeiro, 1977 (in portuguese).
   
   
5. Rodriguez-Amaya DB. Critical review of provitamin A. Determination in plant foods. J Micronutr Anal 1989;5:191- 228.
   
   
6. McCance & Widdowson's. The composition of foods. The Royal Society of Chemistry and Ministry of Agriculture, Fisheries and Food. Cambridge, 1992.
   
   
7. NCR (National Research Council). Recommended Dietary Allowances. National Academic Press, Washington, DC, 10th edition, 1989.
   
   
8. FAO/WHO. Necesidades de vitamina A, hierro, folato y vitamina B12. Estudios FAO Alimentación y Nutrición N° 23. Organización de las Naciones Unidas para la Agricultura y la Alimentación, Roma, 1991.
   
   
9. Roidt L, White E, Goodman GE, Wahl PW, Omenn GS, Rollins B & Karkeck JM. Association of food frequency questionnaire estimates of vitamin A intake with serum vitamin A levels. Am J Epidemiol 1988;128(3):645-654.
   
   
10. ENDEF (Estudo Nacional da Despesa Familiar), Consumo alimentar-Despega das familias, dados preliminares, tabela selecionadas. Secretaria de Planejamento da Presidência da República. Fundação Instituto Brasileiro de Geografia e Estatística, IBGE. Rio de Janeiro, 1978 (in portuguese).
    
    
11. Jarvinen R, Knekt P, Seppaner R, Heinonen M, AAran RK. Dietary determinants of serum beta-carotene and serum retinol. Eur J Clin Nutr 1993;47(1):31-41.
    
    
12. Jacques PF, Sulsky SL, Sadowski JA, Phillips JCC, Rush D, Willett WC. Comparison of micronutrient intake measured by a dietary questionnaire and biochemical indicators of micronutrient status. Am J Clin Nutr 1993;57(2): 182-189.
    
    
13. Coates RJ, Eley JW, Block G, Gunter EW, Sowell AL, Crossman C, Greenber RS. An evaluation of a food frequency questionnaire for assessing dietary- intake of specific carotenoids and vitamin E among low-income black-women. Am J Epidemiol 1991;134(6):658-671.
    
    
14. Scott KJ, Thurnham DI, Hart DJ, Bingham SA, Day K. The correlation between the intake of lutein, lycopene and beta- carotene from vegetables and fruits, and blood plasma concentration in a group of women aged 50-65 years in the UK. Brit J Nutr 1996;75(3):409-418.
    
    
15. Boeing H, Bohlscheidthomas S, Voss S, Schneeweiss S, Wahrendorf J. The relative validity of vitamin intakes derived from a food frequency questionnaire compared to 24-hour recalls and biological measurements: Results from the EPIC pilot study in Germany. Int J Epidemiol 1977;26(suppl):S82- S90.
    
    
16. Christian P, West KP, Khatry Sk, Katz J, Shrestha SR, Pradhan EK, Leclerq SC, pokhrel RP. Night blindness of pregnancy in rural Nepal-nutritional and health risks. Int J Epidemiol 1988;27(2):231-237.