The current management of acute diarrhea is based on oral rehydration salts (ORS) solution for prevention and treatment of dehydration, continued feeding, and antibiotics only for the treatment of cholera and dysentery. Standard ORS solution does not reduce the average duration or severity of acute watery diarrhea or the risk of episode persistence.

Nearly 20% of diarrheal episodes last longer than 1 wk (Sazawal et al. 1995) and 10% longer than 2 wk (Bhan et al. 1989). When children revisit physicians because diarrhea has not ceased within 5-7 d, the latter often prescribe ineffective and potentially harmful anti-diarrheal drugs. In addition, mothers, often on the advice of physicians, change the frequency, amount and consistency of foods offered to the child, which severely curtails dietary intake. The subsequent lack of nutrients may convert marginal into severe malnutrition, delay intestinal epithelial recovery and further increase the duration and severity of diarrhea and malabsorption. The subset of persistent diarrhea episodes, although smaller in number than acute diarrhea, accounts for 23 to 62% of diarrhea-associated deaths in developing countries (Black 1993). Treatment approaches to decrease the average duration and stool output in acute diarrhea and to reduce the risk of episode persistence are therefore needed.

Further, vitamin A deficiency has been found to be associated with increased diarrheal morbidity (Shahid et al. 1988). A significant reduction in the incidence of severe diarrhea was reported in children given 60 mg of vitamin A at 4-mo intervals (Barreto et al. 1994). Vitamin A is absorbed in sufficient amounts during acute diarrhea (Reddy et al. 1986). It is therefore of interest to determine whether these effects of vitamin A treatment appear early enough following administration to be of significant benefit in the treatment of acute diarrhea.

We hypothesized that treatment with 60 mg of vitamin A during acute diarrhea reduces the average duration and severity of the treated episode and the risk of its persistence. This issue was examined in a randomized placebo-controlled field trial.

MATERIALS AND METHODS

Children attending the solitary government health facility in the area with diarrheal duration of 7 d were considered for inclusion in the study if they were between 12 and 60 mo of age and their weight-for-height was 70% of the NCHS median for age. Diarrhea was defined as the passage of three or more loose or watery stools in the 24-h period preceding enrollment.

Of the 1258 children fulfilling these inclusion criteria, 900 were enrolled. The reasons for exclusion of 358 children were as follows: the presence of signs and symptoms of vitamin A deficiency (30; 8.4%), receipt of a large dose of vitamin A in the previous 6 mo (29; 8.1%), the likelihood that the subject would permanently leave the area (157; 43.9%), presence of associated systemic illness (82; 22.9%), prior enrollment into the study (28; 7.8%), weight-for-height <70% of the NCHS median (4; 1.1%), and non-consent (28; 7.8%). Of the 30 excluded children with clinical vitamin A deficiency, 14 had Bitot's spots and the remaining had night blindness alone. Among the 1258 screened subjects, the prevalence of clinical vitamin A deficiency was 1.0% in those aged 23 mo or less, 5.3% in those 24-36 mo of age and 3.2% in children older than 36 mo.

Table 1. Baseline characteristics of enrolled children treated with vitamin A or placebo1 [View Table]

After informed consent was obtained, the child was examined by a physician. Details were sought on the socio-economic status of the family, features of the current illness and feeding practices.

The study was approved by the institutional and World Health Organization (WHO) ethics committees.

At each visit mothers were asked about the frequency and characteristics of stools, fever, vomiting, cough and other related symptoms, and visits to health care providers. Weights to the nearest 100 g and heights (lengths) to the nearest 0.1 cm were measured at enrollment and recovery.

Serum vitamin A was estimated in 40 randomly selected children in each group at baseline and 1 mo later. This sample was sufficient to allow detection of a 20% difference in post-intervention mean serum retinol concentrations between the two treatment groups with 80% power and 95% confidence. Vitamin A concentrations were determined by HPLC by using a reversed-phase column and water-methanol (ratio 5:95) eluant (Bieri et al. 1979).

The field staff was trained in measurement of respiratory rate, temperature, weight, height and hydration status and in giving standard instructions regarding fluids and feeding. Independent checks were made by supervisors for one third of all morbidity visits and half of the measurements obtained by field workers.

Diarrhea and dysentery were treated according to WHO recommendations. Two packets of WHO Oral Rehydration Salts (ORS) were provided to all mothers. Dysentery was treated with nalidixic acid for 5 d.

Student's t test (two-tailed) was used for comparisons of continuous variables between groups; those with skewed distribution were normalized by log transformation, and a two-tailed t test was applied to the transformed data. A paired t test was used to compare pre- and post-intervention serum retinol concentrations. The confidence intervals (CI) for means and difference in means were calculated (Gardner and Altman 1989). Categorical variables were compared using the chi square or Fisher exact test (Armitage and Berry 1987).

Episodes that lasted for 14 or more days post-enrollment were classified as persistent.

RESULTS

The paired difference in mean serum vitamin A concentrations between baseline and 1 mo after treatment was 0.36 ± 0.71 µmol/L (95% CI 0.12 to 0.59; P < 0.001) in the 40 randomly selected children in the vitamin A group and 0.12 ± 0.7 µmol/L (95% CI-0.11 to 0.36; P = 0.136) in the 40 children in the placebo group. Comparison of the means of the change in serum retinol concentration between baseline and 1 mo after supplementation across the two treatment groups did not show significant differences (P = 0.13). Furthermore, the mean serum retinol concentrations 1 mo after supplementation were similar in the vitamin A (1.30 ± 0.44 µmol/L) and placebo (1.17 ± 0.42 µmol/L) groups (P = 0.20).

Subclinical vitamin A deficiency at baseline (serum retinol <0.7 µmol/L) was detected in 26.3% children: 31.6% in the vitamin A group and 21.1% in the placebo group (P = 0.30). The baseline mean serum retinol concentration was 1.0 ± 0.52 µmol/L in breast-fed children and 0.98 ± 0.38 µmol/L in those not breast-fed (P = 0.83).

Table 2. Post-enrollment clinical outcomes in children administered vitamin A or placebo [View Table]

In the analysis on all children aged 1-5 y, the risk of persistent diarrhea was significantly reduced in the vitamin A-treated children (OR 0.30, 95% CI 0.07-0.97). However, the mean diarrheal duration, number of diarrheal stools passed during the episode and the weight changes during the study were similar in the two groups (Table 2).

Undernourished children and those that were not breast-fed may be expected to have a higher prevalence of severe subclinical vitamin A deficiency, so vitamin A may have a greater impact in these subgroups. A standard test of interaction (Pocock 1983) suggested a significant interaction between breast-feeding status and effect of vitamin A supplementation on the mean diarrheal duration (P = 0.01), whereas there was no such interaction for weight-for-height status categorized as 80% and >80% of NCHS median (P = 0.60). The children who received any breast-feeding in the 24-h period prior to enrollment were categorized as breast-fed.

In a secondary analysis, we therefore compared the outcomes in children supplemented with vitamin A or placebo within the subgroups of breast-fed and non-breast-fed children. Among the non-breast-fed children, there was a significant reduction in all the main study outcomes in the vitamin A-treated episodes: 16% in the average diarrheal duration, 27% in the mean stool frequency, and 60% in the proportion of children who passed watery stools. Eight (3.6%) episodes in the placebo and none in the vitamin A group became persistent (Table 3). Among the breast-fed children, there were no significant differences in any of these outcomes between the two groups.

Table 3. Post-enrollment clinical outcomes in non-breast-fed children administered vitamin A or placebo [View Table]

The mean diarrheal stool frequency segregated by study day in the non-breast-fed children is shown in Figure 1. Compared with the placebo group, the vitamin A-treated non-breast-fed children had a 11% reduction in stool frequency on d 1 (P = 0.12), 20% on d 2 (P = 0.03), 16% (P = 0.16) on d 3 and 27% on d 7 (P = 0.01).

To confirm that the favorable effect of vitamin A in non-breast-fed children was not the result of confounding, we performed a multiple linear regression analysis restricted to this subgroup. In this analysis, the outcome variable was diarrheal duration on a log scale (model 1) and the mean number of stools on a log scale (model 2). The explanatory variables were treatment group (placebo or vitamin A), age category (23 or >23 mo), sex (male/female), weight-for-height 80% of NCHS median (no/yes), pre-enrollment duration of diarrhea (3 or >3 d), blood in stools (no/yes), stool frequency in the 24 h before enrollment (median/>median), and consumption of ORS before enrollment (no/yes). All the variables entered in the models were those associated with either of the outcome variables in the univariate analysis at a significance level of P < 0.20, with the exception of sex and pre-enrollment diarrheal duration.

The beneficial effect of vitamin A, after adjustment, remained significant for both mean diarrheal duration (P = 0.03) and mean number of stools during the episode (P = 0.02). A high pre-enrollment 24-h stool frequency was also independently associated with an increase in the mean number of stools passed after supplementation (P = 0.03). No other covariates were significantly associated with either of the outcome variables.

DISCUSSION

Because of the significant interaction between breast-feeding status and vitamin A effect on acute diarrhea, we analyzed results by breast-feeding status. This showed that the effect of vitamin A treatment on acute diarrhea depends on breast-feeding status; vitamin A treatment significantly improves the outcome of the treated diarrheal episodes in the non-breast-fed children, whereas there was no significant effect in those who were breast-fed. In the non-breast-fed children, there was a significant reduction in all the primary outcomes, including stool frequency, days when watery stools were passed and the risk of persistent diarrhea.

These findings are consistent with reports that breast-feeding beyond the first year of life is protective against severe vitamin A deficiency (Cohen et al. 1983, Mahalanabis 1991, Stanton et al. 1986, Tarwatjo et al. 1982, West et al. 1986). The mechanisms underlying the observed effect may be that the correction of subclinical vitamin A deficiency results in a rapid and effective repair of the intestinal epithelium following an acute enteric infection, because of the role of vitamin A in regulating cell division (Semba 1994) or enhancing immune response. Vitamin A has been shown to potentiate the antibody response to a variety of antigens, including rotavirus, E. coli and cholera toxin (Ahmed et al. 1991, Friedman et al. 1991, Wiederman et al. 1993) as well as T cell function (Coutsoudis et al. 1992, Semba 1994).

Furthermore, among several recent studies in Brazil, India, Haiti and Ghana that examined the effect of routine large-dose vitamin A supplementation on subsequent diarrheal morbidity, a convincing reduction in diarrheal morbidity was shown only in the Brazilian study (Barreto et al. 1994, Bhandari et al. 1994, Ghana VAST Study Team 1993, Stansfield et al. 1993). It is noteworthy that in this study, only 13.5% of the enrolled subjects were breast-fed, compared with 47.8% in the Indian study and 58% in the Haitian study.

The observed reduction in the risk of persistent diarrhea is extremely important because the risks of growth faltering and a fatal outcome are high with this disorder. Reduction in the risk of developing persistent diarrhea may be one of the ways by which vitamin A decreases childhood mortality. The relatively low rates of persistent diarrhea in this study compared with earlier studies at the same site may be due to early treatment of dysentery and standard case management of diarrhea. Furthermore, the enrolled children could have been suffering from diarrhea for 7 d prior to the intervention, leading to an underestimate of persistent diarrhea.

The only other trial in which the therapeutic efficacy of vitamin A (60 mg) for acute non-cholera watery diarrhea was examined showed no effect of such treatment on stool output or average diarrheal duration (Henning et al. 1992). However, it is noteworthy that in this study 89% of patients were breast-fed and the sample size of 83 children was sufficient to detect only a 55% difference in the mean stool output between the two groups.

On the other hand, vitamin A administration during acute measles has been shown to reduce the risk of diarrhea and its severity (Hussey and Klein 1990, Ogaro et al. 1993). In the study by Hussey and Klein (1990), children with measles had a significant reduction in complications including pneumonia and diarrhea following vitamin A supplementation; the mean duration of diarrhea was 5.6 d in the treated children and 8.5 d in the controls.

Some limitations of the study need to be emphasized. First, for ethical reasons, all the children were given ORS packets and parents instructed on their use, making it difficult to assess the effect on the risk of developing dehydration. The 27% reduction in the mean stool frequency and the fact that vitamin A-treated children were 60% less likely to have watery diarrhea suggest that vitamin A may reduce stool output and the risk of dehydration in non-breast-fed children. Second, the most significant findings of the study are based on subgroup analysis and need to be confirmed by others.

The effect of vitamin A, although restricted only to the non-breast-fed children, is important because in many developing countries breast-feeding is practiced only during the initial few months, and in other countries breast-feeding rates decline by the end of the first year (Popkin et al. 1982, WHO 1981). In addition, the majority of episodes of severe persistent diarrhea occur in children who are not breast-fed (WHO 1988).

In conclusion, vitamin A treatment during acute diarrhea substantially reduced the episode severity, including the risk of its persistence, in non-breast-fed children. A similar effect was not observed in breast-fed children. These results need to be confirmed in other studies.