点击显示 收起
Division of Gastroenterology and Nutrition The Hospital for Sick Children 555 University Avenue Toronto, ON M5G 1X8 Canada E-mail: stanley.zlotkin{at}sickkids.ca
Dear Sir:
The major criticism by Solomons and Schümann of our recently published study (1) was our lack of a placebo control in the protocol, although their stance on this issue is equivocal, because they also criticize previous studies that included a placebo control. We stand firmly by our decision not to include a placebo control group. We recognize the extensive international debate on this issue (2, 3). In our study of infants in Ghana who had moderately severe anemia, subjects in a placebo group would have received no treatment, which is exactly what they would have received had they not been enrolled in the study. Yet, there is extensive literature on the adverse effects of moderately severe anemia (4). If the current local standard therapy is to do nothing, is it justifiable to include that standard therapy as a study arm? We agree with other ethicists that the standard of providing no less than people are able to get in their own country (which often is nothing) is a standard that is too low (S Benetar, unpublished observations, 2000). We strongly believe that, in clinical research, it is not ethical to offer a "treatment" of doing nothing just because that is what otherwise would be available to the population under study.
Solomons and Schümann make mention of earlier placebo-controlled vitamin A supplementation studies in Northern Ghana by members of our research group, which were conducted between 1988 and 1991 (5, 6). These studies formed part of a second generation of important field trials carried out in several countries that verified the landmark findings of Sommer et al (7). At that time, there was no global policy on universal vitamin A supplementation to children; nor was there a national program in Ghana. The protocols for these trials were reviewed and approved by the relevant ethics committees and the Ministry of Health of Ghana. We believe that it is unethical to withhold treatment of anemia today, and similarly, it would be unethical today to conduct a placebo-controlled trial of vitamin A supplementation in children. This issue begs the larger question of how to identify when the global scientific community has concluded that a specific research question has been adequately answered. In the case of treatment of iron deficiency anemia, we believe that the question has been answered and that further placebo-controlled confirmation studies are both unjustifiable and unethical.
Solomons and Schümann suggest that the biological impact of our not having included a placebo control group would place them at "a loss to assess the attributable efficacy of either form of iron in this study." Statistically speaking, this is true, but is there any biological plausibility to their statistical contention? We believe there is very little. First, the study was of very short duration. What possible mechanisms would lead to significant improvement in anemia in a group receiving no intervention for 2 mo? Because we believe that the primary mechanism causing the anemia was a lack of bioavailable dietary iron, we would have to ask whether a no-intervention group could have rapidly improved their diet with more sources of bioavailable iron, such as meat or poultry. This could have happened if there had been a significant improvement in the economic status of families in the region—for example, that resulting from the sudden discovery of gold or oil deposits. That did not happen, and, to the best of our knowledge, the economic standards of the communities in the region remained static (and very poor). Thus, we doubt that anemia would have improved if a no-intervention group had been included.
Regression to the mean might have accounted for some of the test-retest increment in hemoglobin concentrations, but true regression to the mean is a time-dependent statistical measure. Over the relatively short intervention Solomons and Schümann describe, regression to the mean would have been negligible. They also suggested that the wide biological variation in capillary finger-stick samples (wider than that in venous samples) might have contributed to the regression to the mean. Again, we disagree. Although we acknowledge the wider variation in capillary samples, the variation may result in either an overestimation or an underestimation of hemoglobin concentration (compared with concentrations in venous samples) and thus would not contribute to a regression to the mean (which would imply a bias only to higher hemoglobin values).
There is nothing in the literature to support Solomons and Schümann’s contention that economically disadvantaged infants with anemia due to dietary inadequacy will rapidly improve with no intervention. Thus, although we cannot calculate the attributable change in incidence of anemia in the population we studied, we maintain that the argument against any treatment is fallacious.
Solomons and Schümann question our ability to comment on safety without having included a no-treatment group. They accuse us of "efficacy relativism." We are guilty as charged, but we would argue that we did not include a no-treatment group for the reasons previously outlined and, perhaps more important, that neither ferrous fumarate or sulfate needs the rigorous scrutiny of a "new treatment." Both have been used for decades with documentable side effects such as dark stools, strong metallic taste (ferrous sulfate), staining of teeth (ferrous sulfate drops), and change in gut flora but little else. In sprinkles, ferrous fumarate is coated with a thin layer of soybean lipid to mask the taste of the iron and to prevent changes in the color or taste of the food to which the sprinkles are added. But the form of iron (ferrous fumarate) is not new, and, indeed, in our reported study, given that the fundamental comparison was between the 2 formulations (ferrous sulfate and fumarate), the comparison of diarrhea prevalence was similarly valid.
Solomons and Schümann make the valid point that not all anemias, even in malaria-endemic areas, are due to iron deficiency. We totally agree. Some anemias may be due to blood loss from parasitic infection or to other nutrient deficiencies. One of the advantages of the sprinkles concept is that several micronutrients can be included in the sachet for delivery to the infant. In the current study, we included ascorbic acid to enhance iron absorption. In an ongoing project in Mongolia, we included vitamin D (in addition to iron, ascorbic acid, zinc, vitamin A, and folic acid) in the sachet, because, in Mongolia, anemia and vitamin D deficiency rickets are the 2 most prevalent nutritional problems in children. Thus, we agree with Solomons and Schümann that our study, along with the other studies cited in their letter, "underlies a very important program and policy issue for micronutrient deficiency interventions in developing (and developed) countries...." However, we disagree with their conclusion that "we may be obliged ... to screen and diagnose anemias before assigning treatments...." If their conclusion were given credence, the burden of the human and financial costs incurred would literally put a halt to international anemia intervention programs, which are already going at an unacceptably slow pace. Since 1990, when the World Health Assembly on Children proclaimed that the prevalence of anemia should be lower by the year 2000, it has actually increased (8). Instead of the retrogressive measures suggested by Solomons and Schümann, we believe that there are several innovative interventions on the horizon for anemic infants in malaria-endemic settings, including the Malaria Intermittent Treatment that is being seriously considered by the World Health Organization (9) and the use of sprinkles containing several micronutrients.
Finally, Solomons and Schümann were confused about the description of side effects and adherence to treatment. We originally wrote that "starting 2 wk after baseline and then every 2 wk for the duration of the study, mothers were asked about compliance with treatment over the preceding 7 d, and fieldworkers checked supplies of drops and sachets." We used all of the data from the 4 monitoring visits, and thus the denominator represents the combined data.
REFERENCES