 |
|
Original Article Indian Pediatrics 2002; 39:1099-1118 |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Effect of Vitamin A Supplementation on Childhood Morbidity and Mortality: Critical Review of Indian Studies |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Piyush Gupta, *Abhaya Indrayan
VITAMIN A deficiency is
widely Evidence that has accumulated over the last few years led to belief that improving vitamin A status of children in communities with vitamin A deficiency as a public health problem exerts a measurable positive impact on child mortality and morbidity. A meta-analysis of 10 mortality and 23 morbidity trials(3) conducted till 1993 revealed that improving the vitamin A status of children aged six months to five years reduces mortality rates by about 23%. An important finding was that the effect on mortality was not dependent on very high potency dosing of vitamin A. In contrast to the very clear effect of vitamin A on mortality, no consistent effect was demonstrated on the frequency or prevalence of diarrhea and respiratory infection, though improved vitamin A status did appear to reduce severe morbidity, particularly in children with measles. This meta–analysis included one study each from Tamil Nadu and Hyderabad, conducted in Indian children(4-6). In India, mega dose vitamin A prophylaxis (VAP) was started in 1970 as a program with the specific aim of preventing nutritional blindness. The current policy is to administer vitamin A to all children between 9 months to 3 years. For logistic reasons, VAP has been linked to measles and first booster for DPT immunization(7). The coverage is abysmally low at 30%(8). Universal supplementation of vitamin A to Indian children requires enormous resources and management skills. However, it is noteworthy that even without any significant coverage by the national programs, the prevalence of Vitamin A deficiency in India has declined from 2% in 1975-79 to 0.21% in 1998(8,9). It has also been suggested that mega dose VAP as a measure for control of keratomalacia must be phased out and efforts made to increase the intake of vitamin A through dietary improvement(10). Green leafy vegetables and yellow fruits, freely available in the countryside and well within the economic reach of even the poor, if used judiciously, could control the problem(11). Universal vitamin A supplementation to children has been claimed to bring down the child mortality(3). This claim has been challenged. Massive dose of vitamin A could actually increase child mortality in some situations(11). Therefore, a program of such a magnitude can be undertaken only after convincing evidence is available of substantial benefit in Indian conditions. In this article, we critically examine the methodology of all published randomized controlled trials concerned with linkage of vitamin A supplementation with childhood morbidity and mortality in India. Methods We aimed to examine all published randomized controlled trials of vitamin A supplementation in Indian children for the prevention of death or morbidity from infectious diseases, in particular respiratory and gastrointestinal diseases. To locate primary research data, PubMed was searched independently, using combination of the following key words: vitamin A, morbidity, mortality, randomized controlled trial, and India. In addition, review articles, references of the available meta-analyses, editorial and primary studies were checked to identify other references. Efforts were not made to identify any unpublished trial. The studies were then extracted and the methods section of each study was reviewed with regard to parameters that in our opinion are important for a vitamin A prophylaxis trial in children. These parameters are detailed below:
Results A total of 12 articles (4-6,12-20) fulfilled the inclusion criterion. The manuscript by Biswas et al.(20) could not be retrieved and omitted from the present analysis. Another multicentric study conducted in India, Ghana, and Peru(21) was also not considered, as separate data for Indian children was not available from the published article. Out of 11 articles available for scrutiny, two dealt with reduction of both mortality and morbidity (6,12). In addition, Rahmathullah et al. (4,5) and Agarwal et al. (14,15) published two papers each, studying the impact of vitamin A supplementation on mortality and morbidity separately. Another four articles examined the impact on morbidity alone(13,16-18). A recent study did not directly study the morbidity or mortality following vitamin A supplementation; instead, it concentrated on nasopharyngeal pneumococcal colonization, an indirect determinant of illness or death due to infections with Streptococcus pneumoniae(19). Table I provides a brief summary of the studies included in this review. The following account scrutinizes the methodology of these studies. TABLE I – Summary of Studies Linking Vitamin A with Reduction of Mortality and Morbidity
ARI– acute respiratory infection
Background indicators Eight studies were conducted in the rural settings (5 in Tamilnadu alone), 2 in urban slums (one each in Mumbai and Delhi) and one was hospital based. Only 4 studies provided socio-economic details. None of the study was multicentric or truly represented the entire target population, i.e., the children of India. Subjects Effect of vitamin A supplementation was considered in different age groups ranging from birth to 6 years (Table I). Number of subjects in agewise sub-categories did not conform to that in the general population. Both sexes were represented equally. Pre-existing vitamin A status of study subjects not ascertained in most studies. Design Except one study(12), all others were randomized trials. A placebo was not used in two of the trials. Out of 4 mortality trials, 3 had enrolled more than 15,000 subjects each(4,6,14) but did not calculate or mention the adequacy of sample size and power of these studies. In other studies, where sample size was calculated, it was not sufficient enough to detect differences in sub-categories of age and sex. Intervention Various doses of vitamin A were employed (Table I). None of the studies compared vitamin A supplementation with balanced diet or attempted supplementation through diet alone. Adequacy Out of 11 studies, only 2 trials(13,19) fulfilled the criterion for adequate trial execution and statistical analysis. The first(13) showed no significant impact of vitamin A supplementation on frequency and duration of diarrhea and respiratory tract infection related morbidity. The results of the later(19) can not be generalized as it included only children aged 0-6 months, and was not planned to study a direct relation with mortality or morbidity. None of the mortality trials were considered adequate for execution and analysis. Bias Potential sources of bias existed in all the studies considered for this review. None attempted to identify them and make appropriate statistical corrections. See Appendix for detailed study-wise identifica-tion of bias. The studies quoted in this review were not without their merits. Three of them had a sample size in excess of 15,000. Most were community based trials and published in reputed Journals; and few had been included in meta-analyses on the subject(3,22). Individual study-wise details on each of the parameter (including merits and demerits) are provided in the Appendix. Discussion The issue of reduction in childhood mortality with prophylactic vitamin A supplementation is addressed to in 4 trials on Indian children, out of which 3 have concluded that vitamin A supplementation has a significant impact on bringing down the child mortality. The maximum mortality reduction (78%) was observed in 1-5 year mortality in a trial(12) that was very inadequate in design, quality control, and had the maximum sources of bias; the analysis was inadequate and confounders were not studied. The same trial, however failed to document any decline in the infant mortality rate! Marked reduction in child mortality (54%) was reported in another well conducted trial(4) that supplemented weekly vitamin A at the level of dietary allowances for a year. In the other two trials(6,14) employing mega dose vitamin A supplementation (range 50000 to 200 000 IU) given at 4-6 month intervals, no significant impact was noted on childhood mortality, though the latter study(14) asserted reduction in mortality despite non-significant difference. Nine studies (5-6,12-13,15-19) were examined for possible relationship of vitamin A supplementation and reduction of morbidity. Of these, 5 trials (5-6,12-15) were conducted in otherwise well children. Different supplementation schedules were followed including weekly low dose vitamin A, or megadose vitamin A given 4-6 months apart. Three reports (5,6,13) with relatively better study design reported no significant impact on morbidity; Agarwal et al.(15) reported a mild (10%) reduction in overall morbidity. Kothari(12) also reported some decline in morbidity but important details were lacking, including unvalidated results and conclusions. Two supplementation trials(16,17) were carried out in children with acute diarrhea. The outcome measure in one of these studies(17) was duration of the acute diarrheal episode while the other study(16) aimed to examine the reduction in subsequent morbidity. Vitamin A supplementation was neither found to affect the duration of acute diarrhea, nor did it lower the incidence of subsequent diarrhea and acute respiratory tract infections. In another study(18), morbidity in infancy was not found to be affected by mega dose vitamin A to the mother soon after birth or to her infant at 6 months of age. Coles, et al.(19) tried to link vitamin A supplementation at birth to delayed naso-pharyngeal colonization with pneumococci. Though they did not consider mortality as an outcome measure, it was noted that the number of deaths (amongst the children who dropped out) was equal in both supplemented and placebo group, up to 6 months of age. This fact was not discussed in the study. It is abundantly cleat that on the basis of current studies, no consensus statement can be made. All these trials suffer from methodological flaws, the most important and common are highlighted. Firstly, the sample size, when divided into age-sex groups, nutrition status or health access may not be adequate to detect medically important differences between the treatment and control groups. Secondly, information is lacking on: (i) assured masking till the analysis of data, and (ii) quality control and reliability of the intended procedures and their field imple-mentation. Thirdly, parameters such as diet and health access are not considered as possible confounders and the role of frequent contacts (Hawthorne effect) has not been studied. No attempt has been made to identify the kind of children who could really benefit from the supplementation. Fourthly, multi-variate analysis of data has not been done (except in one study) that would have facilitated the individual net effect of different variables. Potential sources of bias were not identified and corresponding adjustments were not made. Intermediary trend was not analyzed, perhaps due to small sample size. Lastly, none of the studies has addressed the issue of efficacy versus effectiveness, or conducted cost-effectiveness of such inter-vention as against the other measures. All the trials are aimed at only supplementation but none addressed the issue of adequate supply of vitamin A through diet alone, and its impact on childhood morbidity and mortality. The results of the trials done so far are not un-equivocal, particularly findings for mortality and morbidity are not corroborative. We can conclude that there is no definite evidence as yet of substantial benefit of universal vitamin A supplementation to children in India. There is a clear need to undertake a comprehensive trial with adequate sample size and a standardized methodology that could give clear, unbiased, and convincing evidence one way or the other. Since India is a vast country with enormous diversity in baseline characteristics, a multicentric trial with common protocol may provide the final answer. Contributors: AI conceived the idea for this paper and provided the study design. Both the authors were involved in final study design; and collecting, synthesizing and analysing the data. The manuscript was drafted by PG and AI. PG shall stand as guarantor for the paper. Funding: None Competing interests: None stated.
Details of Studies Linking Vitamin A with Reduction of Mortality and Morbidity Studies 1, 2 Rahamatullah et al.(4,5) conducted two trials in villages of Trichy district, Tamil Nadu, and concluded that child mortality reduced on an average by 54% (from 10.5 to 4.8 per 1000) by weekly low dose vitamin A supplementation at the level of dietary allowance. The incidence, severity or duration of diarrhea and respiratory illness was not affected. Mortality trial Background: Drought-prone area. Only 1% received vitamin A supplementation under routine program. Yet, very low U5MR–8.1/1000. Socio-economic details not provided. Subjects: Age 6 to 60 months–increasing from 20.0% in 1-2 years to 22.4% in 4-5 years (contrary to general pattern). Males are 52%, which is fine. Design: Randomized controlled masked trial. Number of Panchayat Unions was three. Criteria of selection not mentioned. As many as 206 clusters formed of variable size (50 to 100 subjects). Unit of randomization was cluster. Total 15,419 children—treatment group 7764 and control group 7655. No mention of adequacy of sample size and power of the study. Intervention: Weekly 8333 IU vitamin A (same as dietary allowances) + 46 mmol vitamin E/mL for one year. Placebo was weekly 46 mmol vitamin E/mL. Proximal measures: Six monthly ocular examination, anthropometric measurement and morbidity history (recall period not mentioned). Finger-prick blood and dietary history taken (presumably six-monthly). Diseases studied but their definition not specified. Mortality (including accidental death) is the main outcome measure. Field work: Steps taken for quality control of data. Pilot study not mentioned. For baseline, equivalence of age-sex, incidence of diarrhea and respiratory diseases, anthropometry, xerophthalmia, U5MR, household economic and hygiene status, and serum retinol level mentioned but no data shown. Health access not studied. Percentage that refused to participate not mentioned. Finger-prick blood and dietary history taken for only 2% random subsample (n = 280). Nearly 12% lost to follow up. Nearly 42% received all the doses (433,000 IU) and 90% received at least 307,000 IU. Children with eye signs continued in the follow-up. Potential sources of bias: (i) Large dose of vitamin A given to children with xerophthalmia (11%) and they continued to be part of the study subjects. (ii) Doses were missed in many children. (iii) Possible large effect of weekly contact. (iv) Recall period for morbidity not mentioned. (v) Possible differentials in nutrition status of children lost to follow-up (12%). (vi) No mention about time lag between baseline assessment and start of the trial. Analysis: Rise in variance (stated as 30%) due to cluster sampling recognized and adjusted. Effect of large dose to children with xerophthalmia ruled out. No analysis done for the effect of missed doses. Contact effect ruled out but substantial difference in mortality in the first 8 weeks not explained. Analysis done to indicate that significant association between treatment and mortality persisted when simultaneously adjusted for age, sex and nutrition status. Extent of decline in mortality not assessed as the baseline level is not mentioned. Cost-effectiveness not studied. Documentation: Data for baseline comparison not provided. Morbidity information is absent. Very few deaths by respiratory diseases not explained. Morbidity trial See preceding Mortality trial for details. Episode of diarrhea and respiratory infection defined. Other morbidity symptoms recorded but not analyzed. Nutritional status cate-gorized as normal, stunted, wasted and stunted+wasted. Detailed analysis of morbidity and growth in children with nutritional status presented. Factors such as health access and diet not considered. Study 3 Vijayraghavan et al.(6) conducted a trial in villages around Hyderabad with six monthly vitamin A doses (maximum 2 doses of 200,000 IU each ) to children aged 1-5 years. The study failed to demonstrate any significant impact on mortality or morbidity. Mortality declined in both treatment and control group with higher number of contacts. Salient features are given below. Background: Villages in backward district. Yet lower child mortality (6/1000) than national average (20/1000). No distribution of vitamin A done in this area. Socioeconomic details not mentioned. Subjects: Age 1 to 5 years—nearly equal in age 0-1, 1-2, 2-3 and 3-4 years but less than half in 4-5 years. Almost equal representation of the two sexes. Design: Double-blind placebo controlled trial. Study area comprised five PHCs (criteria of selection not mentioned). Unit of randomization is village (84) with (presumably) 42 villages in treatment and control group. Of 15,775 children, 7691 are in treatment group and 8084 in control group. No mention of sample size determination and power. Intervention: Vitamin A 200,000 IU six-monthly (max. 2 doses). Placebo is arachis oil. Proximal measures: Mortality noted without cause of death. Disease (diarrhea and respiratory) defined but only frequency considered (severity and duration of sickness not considered). Height and weight taken every six months. Vitamin A status assessed by xerophthalmia (corneal involvement considered severe deficiency). Serum retinol level not measured. For PEM, children categorized by wt-for-ht<80% and 80% but result not mentioned. Diet not studied. Field work: Quality control done by deploying trained staff and supervision. Pilot study not reported. Baseline stated similar for income, age-sex, wt-for-age, birth rate and death rates, but not supported by data. Refusal to participate not mentioned. Mortality and morbidity recorded at three-monthly visits but recall for morbidity was one month (thus many episodes might have been missed). Clinical examination done every six months. First dose given soon after first examination and thus no gap between baseline and start of the trial. Nearly 58% received both the doses in either group, others one or no dose. Lost to follow-up not mentioned. Children with corneal involvement excluded. Source of bias: (i) One-month recall for morbidity whereas the visit was quarterly. (ii) Decoding of treatment and placebo done after the data collection (presumably before data analysis). (iii) Nutritional status differential possible in lost to follow-up. (iv) Unequal children (397 in treatment and 638 in placebo group) received no dose. Analysis: Mortality separately analyzed for children with and without xerophthalmia (no significant association found). Documentation: No result provided on vitamin A supplementation and mortality. Instead substantial part devoted to the association between diseases and xeroph-thalmia (no association found). Study 4 Kothari(12) reported the impact of 200,000 IU vitamin A supplementation on morbidity and mortality in urban slum children of Bombay. There was lowering of 1-5 year mortality rate from 16.8 to 3.64 per 1000 (in 3 years) in the experimental group as compared to 19.75 to 18.9 per 1000 in the control group. There was no lowering of infant mortality rate. The study also observed a significant difference in number of children getting more than four spells of illness per calendar year, between the two groups. Background: Two slum areas, 6 km apart within distant suburbs of Bombay. No concrete data provided on socio-economic status demographic profile, immunization coverage health practices or health access. Subjects: Age at enrollment <1 year; study not planned for sex differentials. Design: No sample size calculations done. No randomization. No mention of masking. Intervention: Consent not obtained. Supplementation given to every alternate infant (<1year) in one slum (n = 195; possible 433). Every alternate child in the other slum (n = 192; possible 416) served as control. Supplementation consisted of vitamin A 200,000 IU, Details on number of doses, frequency of administration and length of administration not specified. No mention of any supplementation/placebo to the control group. Proximal measures: Primary outcomes were mortality and morbidity. Cause of death not ascertained. Episodes (spells) of morbidity not defined. Malnutrition identified as one of the morbidities but magnitude or severity not documented. Morbidities not classified with respect to severity, frequency or duration of each spell. Anthropometric data not collected. Nutrition status not graded. Prevalence of xerophthalmia not assessed. Dietary survey not done. Field work: No quality control tools employed. Details of field methods are lacking. No refusals to participate! Baseline characteristics of the two groups not provided or compared. Follow-up dose at 6 monthly intervals using health interview, health examination and examination of records for 3 years. No follow up/deletion of subjects who have or develop eye signs. Potential sources of bias: (i) Very long recall period, i.e., 6 months. (ii) Observing cause-effect over a very long period of time (3 years) without studying any of the possible confounding factors. (iii) Non-randomized study, baseline characteristics could have been different. (iv) Placebo-effect of supple-mentation not countered. (v) Median age at supplementation and morbidity/nutritional status at the time of supplementation not studied. (vi) Sex distribution in two groups not mentioned. (vii) Differential health access for two groups is possible. (viii) Dietary assessment lacking. (ix) Causes of mortality not analyzed. Factual morbidity data not presented. (x) Possible different nutrition status in two groups. (xi) No mention about specific treatment/inclusion/exclusion of those with xerophthalmia. Analysis: No baseline adjustments done. Details of analysis not provided. Though the follow-up was 6 monthly, the data was not analyzed intermittently. Nutrition, breast feeding, immunization, dietary intake not considered as confounders. No multivariate analysis performed. No details of statistical analysis provided in terms of tests of significance. Statistical validation of differ-ences not done. No attempt at elimination of biases as mentioned above. Interpretation: Important details lacking. No mention of randomization, placebo or masking. Too long a recall (6 months). Statistical methods not adequate. Results and conclusions not validated. Documentation: Presentation of the study is inadequate. Results not generalizable. Confounders not studied. Study 5 Ramakrishnan et al.(13) provided vitamin A supplementation (1 lac IU to <1-year-old and 2 lac IU to 1 year old at four monthly intervals, thrice a year), in children 6-36 months, from villages in Tamil Nadu. Supplementation did not appear to reduce common morbidity in children who have mild to moderate vitamin A deficiency. Background: The trial was carried out in villages where Growth Monitoring Research project was going on for two years. Thus, access to health care was good and immunization more than 90%. All children older than 1 year routinely dewormed. Socioeconomic status of the families also assessed. Subjects: Subjects are children of age 6 to 36 months with nearly equal males and females. Design: Randomized double blind placebo control intervention trial. Unit of randomization is child. Treatment group finally had 309 children and control group 274. No mentioned about adequacy of sample size. Power calculated post-study showed adequacy to detect 25% reduction in morbidity. Intervention: Vitamin A 100,000 IU to children less than 1 year and 200,000 IU to children 1 year and above three times four months apart. Placebo not described. Proximal measures: Mortality was not considered. Morbidity (particularly diarrhea and respiratory illness) assessed in terms of percent time ill, incidence and mean duration per episode. Weekly visits done and recall time was one week only. Episode of different disease defined. Height, weight, and midarm circumference measured at the beginning and at the end of the study. Ophthalmic signs of vitamin A deficiency and serum retinol level assessed at the beginning and at the end. Dietary intake of b-carotene and preformed vitamin A estimated by using a quantitative food frequency questionnaire. Stunting, under weight and wasting defined. Growth monitored in half the villages. Field work: Morbidity data collected by trained village nutrition workers on weekly basis. Clinical criteria could not be included. Forms routinely checked by supervisors. Ophthalmic signs assessed by trained ophthalmologist. Pilot study not mentioned. Percentage that refused to participate not mentioned. Baseline characteristics of the children (age, sex, ht-for-age, wt-for-age, serum retinol, vitamin A intake, bottle feeding, birth order and immunization) compared between treatment and control but not the prevalence of xerophthalmia. Socioeconomic status of the parents also compared. Follow-up was for one year. Serum retinol level assessed only for 366 (63%) subjects. Children with signs of xerophthalmia, severe malnutrition (defined) and serious illness (not defined) excluded (55 children). Out of 715, only 583 (82%) used for analysis–others (including 55 just mentioned) lost due to migration, death and inadequate morbidity data. Potential sources of bias: (i) Effect due to weekly contact. (ii) Children lost (18%) not examined for baseline characteristics. (iii) Treatment group has substantially more females than males and maternal education more in treatment group (though not statistically significant at a = 0.05). (iv) Serum retinol level available only for 63% of this truncated group. Analysis: Adjustment for possible bias described above not done. Infectious disease load analysed for diversity, frequency and duration but not for severity. Analysis done by sex, growth monitoring, nutrition status (serum retinol level and degree of stunting, wasting and under-weight). Multivariate analysis also done to adjust for factors as age, sex, nutrition status, immunization status, economic status, vitamin A intake and mother’s education. Interpretation: Results seem internally consistent. Since multivariate analysis was also done, higher error due to multiple tests is ruled out. Power calculated post-study. Documentation: Does not discuss mortality but the study looks complete for morbidity. Studies 6, 7 Agarwal et al.(14), in a trial in villages of Varanasi, concluded that vitamin A administration to children under six years protects against overall mortality, in particular deaths due to gastroenteritis and severe malnutrition. In another trial(15), they demonstrated that vitamin A supplementation for one year resulted in 10% overall reduction in morbidity in children of 1-72 months from villages of Varanasi. Mortality trial Background: Villages in Varanasi district, but less than 15% belonged to lower socioeconomic class. No mention of health access in the selected villages. Socioeconomic data provided. Subjects: Children one month to six years of age but 0-1 nearly twice as many in the treatment group as in 1-2 year age-group – a feature not present in the control group. Also, children of age 5-6 years are more than of 4-5 years. M-F ratio 1000:810 in both the groups. Design: Randomized double blind trial. Unit of randomization is subcenter. By mistake 12 of subcenters (instead of 8) received treatment and only 4 received placebo. Children in treatment group are 9987 and in placebo group 5260 (not in the expected ratio of 3 : 1 as for subcenter). No mention of sample adequacy, size and power. Second phase was one time measurement after one year of withdrawal of intervention. Intervention: Treatment group received 50,000 IU vitamin A plus 10 IU vitamin E if child is 1-6 months and 100,000 IU vitamin A plus 10 IU vitamin E if child is 7-72 months, every 4 months for 12 months. Placebo is only vitamin E. Proximal measures: Primary outcome is mortality. Cause of death also assessed. Morbidity also studied but details not provided (see possibly morbidity trial below). Nutritional status graded on the basis of wt-for-age. Immunization and xerophthalmia status recorded. Serum retinol level and diet not assessed. Field work: No mention of training or supervision as also of pilot study. Baseline age-sex and nutrition stated to be similar without statistical test. Significantly high mortality in the experimental group than control groups at baseline. Xerophthalmia 3.6% in treatment and 2.4% in control group. None refused to participate yet nearly 15% received no dose. No mention of how many received all the doses. Potential sources of bias: (i) Contact effect not considered. (ii) Differential follow-up in terms of doses administered (nearly 10% of treatment group and 20% of control group received no dose). (iii) Differential mortality at baseline. (iv) Unusual predominance of infants in the subjects. (v) Possible differential health access. Analysis: Mortality analyzed by immuniza-tion and nutrition status including for xerophthalmia. Cause of death also analyzed. No adjustment done for bias mentioned above. Disease load not considered. Reduction in mortality asserted despite non-significant difference. Most differences were not significant anyway but multiple statistical tests used on one group at a time has a risk of showing unreal significance. Interpretation: Role of confounding factors such as frequent contacts and differential in base line not discussed. Statistical signifi-cance achieved on the basis of increased mortality in control group and decreased in the treatment group, each of which may not be significant. Documentation: No result provided for morbidity. Four-monthly trend in mortality not studied. Morbidity trial Background: Villages in Varanasi district with IMR 100 and low birth weight deliveries 26%, socioeconomic details provided. Nearly 25% received immunizations. Subjects: Children of age 1 month to 6 years. Age distribution not provided. Males and females nearly equal. Design: Randomized double blind controlled trial. Six villages randomly selected out of 112 in a PHC area. Three each randomly allocated to the treatment and control group. Number of children 1665 in the treatment group and 1440 in the control group. No mention of adequacy of sample size and power to detect strategically important difference. Intervention: Treatment group received 50,000 IU vitamin A plus 10 IU vitamin E if child is 1-6 months and 100,000 IU plus 10 IU vitamin E of child is 7-72 months, every 4 months for 12 months. Placebo only vitamin E. Proximal measures: Nutritional categories by Gomez classification (weight-for-age). Morbidity assessed in terms of sickness days, episodes per child, duration of episodes, etc. Disease included measles, respiratory infection, otitis media and skin infections. Xerophthalmia recorded at baseline but no mention of its status at the end of the trial. Serum retinol level not measured. Diet not assessed. Field work: No mention of training or supervision or a pilot study. Social factors (literacy, family size, access to potable water) not different in the study and control group. Xerophthalmia slightly higher in control group (3.2%) compared to treatment group (2.1%). Nearly 13.5% children of control and 18.5% of treatment group could not be contacted. No mention of how many received all the doses and how many only one or two doses. Follow-up 2-weekly for morbidity and four-monthly for drug administration. Children with xerophthalmia administered 200,000 IU and excluded from analysis. Potential sources of bias: (i) Differential follow-up in terms of doses administered. (ii) Baseline vitamin A status not considered. (iii) Increase in morbidity in control group. (iv) Possible diet differential. Analysis: Overall morbidity analyzed by age, sex, immunization status, nutrition status and social factors (type of family, family size, source of drinking water, literacy) in univariate set-up. Multivariate analysis not done. Disease load assessed for diversity, frequency and duration but not severity. Source of bias due to differential doses not considered. Age wise analysis of individual diseases not done. Interpretation: Too many statistical tests at 5% level that could enormously increase total error. Yet, many results provided without statistical tests. Role of 2-weekly contact for morbidity assessment not ruled out. Documentation: Mortality not studied (see mortality trial above). Four-monthly trend not provided. Effectiveness and cost-effective-ness not considered. Study 8 Bhandari et al.(16) studied the impact of single dose vitamin A 200,000 IU on morbidity from acute respiratory tract infections and diarrhea in children with acute diarrhea in an urban slum of New Delhi. Results were consistent with a lack of impact on acute lower respiratory tract related morbidity or incidence of diarrhea. Authors noted a 36% reduction of diarrhea with fever in vitamin A supplemented children of age more than 23 months. Background: Urban slum. Socioeconomic background provided. Immunization coverage for 3 doses of DPT 74 %. Health access facility average. No distribution of vitamin A in preceding three years. Representativeness of the target population questionable as only children with acute diarrhea were enrolled. Subjects: Age 12 months to 60 months. Approximately half (49.4%) in 12 months-23 months. Males 52%. Only those with acute diarrhea and weight for height 70% of NCHS 50th percentile enrolled (n = 900), out of possible 1258. Those with clinical vitamin. A deficiency (n = 120) excluded. Maximum prevalence of clinical vitamin A deficiency seen in 24-36 months age group (5.3%). Design: Double blind placebo controlled trial. Unit of randomization is child. Treatment group finally had 422 children and control group 420. Sample size calculations provided. The trial size was adequate to detect a 25% reduction in the incidence and prevalence of diarrhea but not to detect a similar impact on the incidence of pneumonia; with a power of 90%. No sample size calculation or post-hoc analysis for age subgroup categories (£23 months vs ³24 months) that the authors have used. Intervention: Single dose vitamin A 200,000 IU. Placebo not described. Proximal measures: Mortality was not considered. Morbidity assessed in terms of incidence and mean daily prevalence of acute respiratory tract infection and diarrhea. Episodes of different diseases defined both in terms of onset and recovery. Severity of diarrhea measured in terms of presence/absence of fever, but not graded for dehydration. Severity of ALRI measured only in terms of lower chest indrawing, and/or presence of pneumonia. Visits made every three days. Weight and length/ height measured at enrolment (data provided) and at the end of the study (data not provided/analyzed). Those with xerophthalmia excluded at the start but no mention of those who developed xerophthalmia during the study. Serum vitamin A level estimated in 40 (5%) randomly selected children in each group, before and a month after supplementation. Dietary intake not assessed. Field work. Informed consent taken. Household visits made by a trained field assistant every 3 days. Reassessment made by supervisors in one-third to one-half of recalls/measurements. No mention of laboratory standardization, any manual of instruction to field workers, or a pilot study. Refusal to participate 7.8%. Baseline socio-economic and clinical characteristics of children compared and found not different. No comparison done for dietary intakes. Follow up done for 90 days after recovery. Out of 900 enrolled, 842 (93.5%) available for final analysis – others lost due to non-availability (n = 46), withdrawl of consent (n = 5), and severe illness (n = 7) including tuberculosis, cardiac disease or severe anamia. Serum vitamin A levels available for only 80 subjects. No mention of those who developed eye signs of VAD during the study. Potential sources of bias: (i) Children lost (6.5%) not examined for baseline characteristics. (ii) Contact effect not considered. (iii) Unusual prominence of <2 year olds (reflects the age preference for diarrhea?) (iv) Severity of initial enrollment diarrheal episode not quantified in terms of associated dehydration. (v) Characterization of severity of diarrheal or respiratory illness not proper. (vi) Duration of study and season not mentioned. (vii) No mention of any mortality in the study subjects. Analysis: Initial socio-economic and clinical characteristics in the two groups shown not different. Serum vitamin A concentration in the two groups found not different. Relative risk of ARIs not different from one. Adjustment for possible bias described above not done. Seasonality not considered. Impact of breast feeding, weaning and maternal nutrition not studied. Characteristics of those who showed improvement not delineated. Multivariate analysis not done. Interpretation: Adequate. Documentation. Does not discuss long-term follow up beyond 3 months. Generalizability is inadequate since only those with acute diarrhea considered and clinical vitamin A deficiency not included. Study 9 This is another hospital-based study that administered 100,000 IU vitamin A to children with acute diarrhea(17). No long-term outcome was seen and the impact of supplementation was observed on the duration of acute diarrheal episode. Vitamin A supplementation did not significantly reduce the duration of a diarrhea episode except in those with pre-existing vitamin A deficiency and associated malnutrition. Background: Hospital based study in subjects reporting to a tertiary care center for treatment of diarrhea Subjects: Age 6 months to 5 years of either sex. Mean age approximately one year. No data provided for different age categories. Subjects not classified by sex. Design: Randomized controlled trial. Method stated as simple random sampling. No placebo used. Not masked. Justification for sample size not mentioned. No stratification done for age-sex but done for socio-economic status. Prior equivalence of case and control group is plausible. Intervention: Single dose vitamin A given orally to 108 cases with acute diarrhea (100,000 IU to the others). An equal number of controls enrolled from the same clinic. (Children with xerophthalmia in control group denied treatment till the diarrhea episode was over, which could have compromised ethics.) Proximal measures: Mortality and long term morbidity not studied. Effect of vitamin A supplementation observed only on an acute episode of diarrhea. Outcome measure was mean duration of diarrhea. Onset of diarrhea not defined though termination is defined. Xerophthalmia was noticed in 11.1% of cases and 12.9% of controls and they continued to be the part of the study. Serum retinol level not studied but conjunctival impression cytology was possible in around 88% of enrolled subjects. Diet was not assessed. Field work: Hospital based study. No mention of quality control, or pilot study. Method of selection of cases (only 108 cases of diarrhea enrolled in more than 9 months in a tertiary care center!) and controls not specified. Duration of maximum follow up not specified. Frequency and method of follow up are also not specified. Refusal to participate not mentioned (probably because no consent was obtained). Cases and controls stated not different for age-sex distribution, feeding pattern, socio-economic status, nutritional status, stool frequency, hydration, associated symptoms but no data provided to substantiate this. Immunization status, duration of exclusive breast feeding, time of weaning, dietary intake, health practices, water supply, housing, etc. not considered. Loss in follow up not reported for any of the cases or controls! Potential sources of bias: (i) Selection of cases and controls not specified. (ii) Cases and controls not studied for several parameters such as immunization status, duration of exclusive breast feeding, weaning time and foods, dietary intake, health practices, water supply, housing that affect vitamin A status or onset termination of diarrhea. (iii) Contact effect not considered. (iv) Period of recall not mentioned. (v) Children with xerophthalmia continued to be part of the study. Analysis: Adjustments for baseline imbalances not needed as none was detected. Power of the study to detect clinically important difference not studied. Impact of breast feeding-weaning not studied. No multivariate analysis done. Interpretation: Conclusion that vitamin A supplementation has beneficial effect in cases with pre-existing VAD is fine with univariate analysis but its validity under multivariate analysis is not known when factors such as age, socioeconomic status, weaning are also considered. Documentation: Restricted only to diarrhea episodes, and morbidity, mortality, retinol level, and xerophthalmia not studied. Study 10 Venkatarao et al.(18) studied the impact of 300, 000 IU vitamin A to the mother soon after delivery and to the infant (200, 000 IU) at six months on morbidity in infancy in a field trial in rural Tamil Nadu. Authors concluded that prophylactic administration of vitamin A megadose to mother soon after delivery and to the infant at six months do not have any beneficial impact on the incidence of diarrhea and ARI in infancy. Background: 51 villages in Tamil Nadu, No mention of health access to inhabitants. Socio-economic details partially mentioned. Immunization coverage 75%, 58%, 83% and 56% for BCG, DPT, OPV and measles respectively. No mention of routine/previous vitamin A distribution in the area. Subjects: A newly delivered mother and her infant formed a pair of subjects. Newly born child followed up till 1 year of age. Equal sex representation. Only 909 (68.1%) pairs enrolled out of possible 1335 child births. Design: Double-blind placebo controlled trial. Unit of randomization was the mother-infant pair. Sample size adequate to detect a 10% difference in the incidence of diarrhea/ARI with a power of 90% (post-hoc calculations); but not analyzed for two age categories, i.e., 0-6 and 6-12 months. Intervention: Both mother and infant received vitamin A (300,000 IU for mothers and 200,000 IU for children) in 311 instances (AA); mother received vitamin A but infant received placebo in 301 instances (AP); and both mother and infant received placebo (PP) in 297 instances. Mother given supplementa-tion between 7-14 days after delivery and the infant received vitamin A or placebo at 6-6½ months of age. Consent obtained from village leaders but not from individual mothers. Placebo to mothers was 600 mg of vitamin E and to infants was 2 mL of sesame oil. Proximal measures: Mortality and cause of death assessed. Morbidity (diarrhea and ARI) assessed in terms of incidence, frequency of episodes and number of infected days. Episodes of diarrhea and ARI defined. Severity definition not appropriate, i.e., all episodes of lower respiratory tract infection (assumed in children with difficulty in breathing for <30 days in children <2 months old, and associated with cough in those ³2 month old) were equated with those requiring hospitalization. Criteria for hospitalization not specified. Similarly episodes of diarrhea associated with vomiting were considered severe but not graded as per the hydration status. Management of an episode not standardized or clarified in the manuscript (who treated, where and how?). Weight recorded at birth at 6-6½ months and one year. Nutrition status not studied on individual basis. Serum retinol levels not obtained either in the mother or infant. Dietary intake of neither the mother nor infant studied. Field work: Field workers trained intensively and procedures standardized during field testing. No mention of instruction manual. No standardization of treatment protocol for morbidities. Supervision initially in 10% random sample and subsequently by spot checking once a week (number or percentage not mentioned). Not preceded by a pilot study. Out of 1335 possible enrollment, 909 (68.1%) enrolled. No reason provided for 135 mother-infant pair exclusions. Of the 909 randomized, final analysis was available for 233, 228 and 228 pairs in three groups respectively. Follow up loss 24.2%. Of 220 drop outs, there were 20 deaths, 13 medical exclusion, 79 migrations and 23 miscellaneous reasons. No reason provided for exclusion of 85 (10.7%) of randomized subjects. There was no refusal to participate because the consent was not asked for! Baseline characteristics (sex, literacy, maternal age, parity, family size, housing, immunization, breast feeding habits, neonatal weight) were not different in three groups; however data not provided. Side effects of vitamin A monitored. Potential sources of bias: (i) Maternal anthropometry, illnesses, hemoglobin, albumin and retinol level not studied. (ii) Newborns were included irrespective of their weight, gestation and appropriateness for gestation. (iii) Exclusive breast feeding rate was not studied or compared in three groups. (iv) Age of weaning, detailed dietary history not obtained. (v) Infants not classified as per their nutritional status. (vi) Weight gain during second half of 1st year of life was only 1.0 kg. (vii) High unaccounted follow up loss, both before and after randomization. (viii) Recall period approximately two weeks. (ix) Possible differential management of morbidities. (x) Children with clinical xerophthalmia continued to be part of the study, not detected separately. Analysis: No adjustments for above biases. No Intention to Treat analysis done. No analysis done for possible confounders such as maternal nutrition, birth weight, gestation, exclusive breast feeding, weaning, diet, serum retinol, and nutritional status. No multivariate analysis done. Analysis provided separately for 0-6 and 6-12 months ages and compared for all outcome variables. Interpretation: Though not significant, a higher incidence of diarrhea was observed in vitamin A supplemented groups. Similar trend was seen with the number of diarrheal episodes and ARI, and illness days, for 0-6 month age analysis. In 6-12 months children also the incidence of diarrhea, mean number of diarrheal episodes, median days with diarrhea was lesser in the placebo group, though statistically not significant. This was not discussed. Documentation: Not comprehensive. Sample size for mortality analysis is incomplete. Retinol levels not considered. Study 11 Coles et al.(19) studied the impact of two doses of 7000 micrograms retinol equivalent doses of vitamin A at birth in reducing the pneumococcal colonization in infants from a rural area in Tamil Nadu with endemic vitamin A deficiency. The risk of colonization among infants aged 4 months who were not colonized by age 2 months was significantly reduced in vitamin A supplemented group. The odds of colonization were 27% lower in the supplemented group. No difference was detected in the prevalence of invasive serotypes. The risk of colonization with penicillin resistant isolates was 74% lower in the vitamin A group than placebo group at 2 months of age. However, the preva- lence of penicillin resistant isolates was only 4%. Background: Natham in rural Tamil Nadu is endemic for vitamin A deficiency and has a high incidence of ARI. No mention of any routine vitamin A supplementation programs. Socio-economic indicators provided. No mention of immunization coverage, health access and health practices. Subjects: 539 infants randomized at birth and followed up at 2 monthly interval up to the age of six months. Slight male preponderance (56.6%). Design: Double blind placebo controlled trial. Unit of randomization was mother, stratified by cluster and blocked within geographical cluster area. Sample size calculated, though the power to detect differences between treatment groups was relatively low, sample size was inadequate to assess the impact of vitamin A supplementation on prevalence of penicillin-resistance strains (due to low prevalence – nearly 4% in this population). Masking appears adequate and coding remained with a third party. Intervention: Newborn received either two doses of 7000 mg retinol equivalents of oil soluble vitamin A or placebo within 48 hours of birth. Verbal consent was obtained from the subjects. Nature of placebo not mentioned. Proximal measures: Mortality data available for drop-outs. Number of deaths was equal (seven each) in both the groups. Morbidity data not provided. Primary outcome of interest was the prevalence of pneumococcal nasopharyngeal carriage. Secondary outcome of interest were (i) prevalence of invasive serotypes of pneumococci, and (ii) prevalence of penicillin resistant pneumococci. Growth, development, nutrition status, dietary intake not considered nor recorded. Field work: Quality control, laboratory standardization maintained. Reliability of data ascertained. Treatment groups not different at baseline for sex, birth weight, siblings, colostrum feeding, maternal night blindness, cooking fuel, property, maternal education and housing. Treatment groups comparable at baseline with respect to known risk factors for ARI but neither the factors specified, nor the data presented. Follow up was done at 2 (n = 464), 4 (n = 406), and 6 (n = 359) months after birth. Refusal to participate mentioned. All losses to follow-up were accounted for. Morbidity assessed at every 15 days but data not provided. Potential sources of bias: Treatment and control group not compared with respect to maternal nutrition, maternal vitamin A status, and infants retinol level, mucosal immunity and secretory IgA levels, that could have affected the prevalence of colonization. In addition, exclusive breast feeding and weaning methods/age not compared. Analysis: Appropriate adjustment for baseline imbalances made. Analysis done by Intention to Treat. Bias as highlighted above not taken care of. Multivariate analysis carried out. Interpretation: Adequate within the constraints mentioned above. Documentation: Mortality not different between groups. Morbidity, growth data, retinol level not provided. Colonization studied only upto the age of 6 months.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
References | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
