Contents - Previous - Next


de Onis et al. reviewed evidence from randomized controlled trials (RCTs) on the effectiveness of nutritional interventions in the prevention of IUGR. Several discussants expressed the view that RCTs should not be considered the only source of information on the effectiveness of interventions. Results of other types of studies can provide useful information too, especially for hypothesis generation, and it is obvious that all promising interventions have not been adequately tested in RCTs yet. The main problem with observational studies is that they tend to be difficult to interpret, because so many potentially growth-limiting factors tend to be correlated, but they are useful for suggesting what interventions might be worthwhile testing in RCTs that allow inferences on causation.

RCT databases like the Cochrane database which was the main source of information for this paper, have a number of advantages compared to other, conventional kinds of reviews: they have a set of clearly stated criteria for inclusion and exclusion of studies, and correspondence between reviewer(s) and database users on the quality of trials becomes integral part of the database. Evaluation thus becomes a continuous, open and transparent process and is not just taking place at a specific point in time in the mind of one reviewer. RTC databases have the additional advantage of being more appropriate for the examination of moderate effects. In fact, they are most appropriate where outcomes are related to multiple causes and where the expected intervention effect is likely to be moderate. Both of these characteristics apply to interventions to reduce the risk of IUGR. Under field conditions successful interventions usually lead only to moderate (50-150 g) increases in birthweight.

RCTs are usually large and sometimes costly studies; most of them have, therefore, been carried out in high-income countries. Findings tend to be population-specific and can therefore not be easily generalized to conditions prevailing in developing countries. RCTs also tend to be designed to gain medical knowledge, i.e., they are geared to the evaluation of individual-level interventions; they do not necessarily lend themselves easily to making public health policy decisions in developing countries. There is therefore an urgent need for more RCTs in developing country settings.

Conclusions drawn from examining RTC databases could be subject to publication bias. Trials with positive outcomes are more likely to be published than trials with negative outcomes; the overall conclusions suggested by reviews could therefore tend to be overoptimistic.

Nutrient supplements have been tested in populations with a low prevalence of deficiency of the nutrient that was supplemented. Interventions can only be effective if they are targeted at populations that are appropriate, i.e., at considerable risk. In many settings, IUGR is likely to be the outcome of a multifactorial process and it seems worthwhile to see if combining several interventions can improve the outcome. Once the effectiveness of an intervention has been demonstrated in RCTs, feasibility also has to be considered before recommendations are made; interventions need to be cost-effective and sustainable.

Contents - Previous - Next