Volume 118 Number 6
withdrawal is rare enough not to be considered a potential source of selection bias. Additionally, one of the reasons we used a 6-month treatment minimum was to avoid the inclusion of children with brief MTX exposure, which might bias one toward falsely low rates of toxic reaction. Feldman and Silverman also note a lack of "hard indicators" for efficacy assessment of MTX therapy. Although more laboratory information might be useful, the variables selected were chosen because they were available for every patient. Further, we are not aware of any study showing leukocyte counts as a reliable method of follow-up for JRA. The use of skeletal radiographs for monitoring treatment efficacy in adults with rheumatoid arthritis has recently been challenged. 7 It is widely recognized that a controlled, blind-design study will reduce the possibilities of bias in data collection and generation. As stated, our study was uncontrolled and retrospective. The choice of an independent observer for medical records review and the use of three visits during 2 months to average findings were employed to minimize some of the biases known to be associated with retrospective studies. The same response is pertinent to the concerns of Feldman and Silverman about disease variability, compliance, and placebo effect as sources of type I error. This group of patients was relatively sick and had multiple prior treatment failures; spontaneous remission or improvement would have been less likely. Uncontrolled studies contain potential and actual biases, but they often also realistically reflect problems faced by physicians on a daily basis. Carlos D. Rose, MD Bernhard H. Singsen, MD Division o f Rheumatology Alfred L duPont Institute Wilmington, DE 19899
REFERENCES
1. Giannini EH, Brewer EJ. Methotrexate (MTX) in the treatment of recalcitrant JRA: results of the double-blind, placebo (P) controlled randomized trial [Abstract]. Arthritis Rheum 1989;32(4; suppl):S82. 2. Tugwell P, Bennett K, Gent M. Methotrexate in rheumatoid arthritis indications, eontraindications, efficacy, and safety. Ann Intern Med 1987;107:358-66. 3. Williams H J, Willkens SF, Samuelson CO, et al. Comparison of low-dose oral methotrexate and placebo in the treatment of rheumatoid arthritis. Arthritis Rheum 1985;28:72130. 4. Truckenbrodt H, Hafner R. Methotrexate therapy in juvenile rheumatoid arthritis: a retrospective study. Arthritis Rheum 1986;29:801-7. 5. Wallace CA, Bleyer WA, Sherry DD, et al. Toxicity and serum levels of methotrexate in children with juvenile rheumatoid arthritis. Arthritis Rheum 1989;32:677-81. 6. Danao T, Steinbrunner J, Medendorp SV, et al. Methotrexate (MTX) in juvenile rheumatoid arthritis [Abstract]. Arthritis Rheum 1989;32:$28. 7. Brower AC. Use of the radiograph to measure the course of rheumatoid arthritis: the gold standard versus fool's gold. Arthritis Rheum 1990;33:316-24.
E d i t o r i a l correspondence
993
Effect of lactase on infantile colic To the Editor: A recent article by Miller et al., "Effect of Yeast Lactase Enzyme on 'Colic' in Infants Fed Human Milk," (J PEDIATR 1990; 117:261-3), came to the conclusion that the lactose content of human milk was not the culprit in colicky babies. This is contrary to our experience. From March 1987 to April 1988, eleven infants with colic participated in a double-blind, prospective, crossover study in which lactase (i n the form of LactAid) or placebo was orally administered for 5 days each. The laetase was given before the initiation of breast-feeding and again in mid feeding. One child was lost to follow-up; seven of the remaining 10 infants responded to the active enzyme. Four of those seven had such a marked and reproducible response that the mothers insisted on additional treatment after the 10-daY study. Although we were not able to do breath hydrogen tests, three of the infants who had a small amount of reducing substance in the stool before the test had a clinical response to the lactase. Our data support the concept that in some breast-fed babies, lactose intolerance is the cause of colicky behavior. The manner in which the lactase was administered--that is, before and again during breast-feeding--seems to be the only difference between our study and Dr. Miller's study; he gave the enzyme at the start of a feeding only. Although Dr Miller's dosage was based on an in vitro test demonstrating 75% hydrolyzation by 25 minutes, the dosage might have been less than needed, as demonstrated by the lack of decrease in the breath hydrogen levels. Harry F. Laws II, Colonel, USAF, M C Pediatrician, 554th Medical Group Nellis AFB, N V 89191-5300
Reply To the Editor: Dr. Laws suggests that lactase may be effective if administered before and during breast-feeding but not when the enzyme is given within 5 minutes of the start of a feeding. It is possible that giving yeast lactase, which is unstable at acid pH, within 5 minutes of the start of a feeding results in inactivation of the enzyme in the stomach. However, a recent study by Moore, 1 involving 39 breast-fed infants who were given lactase toward the end of each feeding, also failed to demonstrate an effect of lactase on the duration of crying and fussing or breath H2 levels. Eight formula-fed infants were also studied. These infants were fed either a lactose-hydrolyzed formula (LHF) or a lactose-containing formula (LCF) for 1 week and then changed to the other formula for a further week. There was no significant difference in the duration of crying and fussing between the babies who who were given LHF and those given LCF. No reduction in breath H2 excretion was observed in the babies receiving LHF. Although reducing lactose intake may not be an effective treatment for colic, it may be prophylactic. We followed 28 healthy infants who were fed either an LHF or an LCF from the first week
withdrawal is rare enough not to be considered a potential source of selection bias. Additionally, one of the reasons we used a 6-month treatment minimum was to avoid the inclusion of children with brief MTX exposure, which might bias one toward falsely low rates of toxic reaction. Feldman and Silverman also note a lack of "hard indicators" for efficacy assessment of MTX therapy. Although more laboratory information might be useful, the variables selected were chosen because they were available for every patient. Further, we are not aware of any study showing leukocyte counts as a reliable method of follow-up for JRA. The use of skeletal radiographs for monitoring treatment efficacy in adults with rheumatoid arthritis has recently been challenged. 7 It is widely recognized that a controlled, blind-design study will reduce the possibilities of bias in data collection and generation. As stated, our study was uncontrolled and retrospective. The choice of an independent observer for medical records review and the use of three visits during 2 months to average findings were employed to minimize some of the biases known to be associated with retrospective studies. The same response is pertinent to the concerns of Feldman and Silverman about disease variability, compliance, and placebo effect as sources of type I error. This group of patients was relatively sick and had multiple prior treatment failures; spontaneous remission or improvement would have been less likely. Uncontrolled studies contain potential and actual biases, but they often also realistically reflect problems faced by physicians on a daily basis. Carlos D. Rose, MD Bernhard H. Singsen, MD Division o f Rheumatology Alfred L duPont Institute Wilmington, DE 19899
REFERENCES
1. Giannini EH, Brewer EJ. Methotrexate (MTX) in the treatment of recalcitrant JRA: results of the double-blind, placebo (P) controlled randomized trial [Abstract]. Arthritis Rheum 1989;32(4; suppl):S82. 2. Tugwell P, Bennett K, Gent M. Methotrexate in rheumatoid arthritis indications, eontraindications, efficacy, and safety. Ann Intern Med 1987;107:358-66. 3. Williams H J, Willkens SF, Samuelson CO, et al. Comparison of low-dose oral methotrexate and placebo in the treatment of rheumatoid arthritis. Arthritis Rheum 1985;28:72130. 4. Truckenbrodt H, Hafner R. Methotrexate therapy in juvenile rheumatoid arthritis: a retrospective study. Arthritis Rheum 1986;29:801-7. 5. Wallace CA, Bleyer WA, Sherry DD, et al. Toxicity and serum levels of methotrexate in children with juvenile rheumatoid arthritis. Arthritis Rheum 1989;32:677-81. 6. Danao T, Steinbrunner J, Medendorp SV, et al. Methotrexate (MTX) in juvenile rheumatoid arthritis [Abstract]. Arthritis Rheum 1989;32:$28. 7. Brower AC. Use of the radiograph to measure the course of rheumatoid arthritis: the gold standard versus fool's gold. Arthritis Rheum 1990;33:316-24.
E d i t o r i a l correspondence
993
Effect of lactase on infantile colic To the Editor: A recent article by Miller et al., "Effect of Yeast Lactase Enzyme on 'Colic' in Infants Fed Human Milk," (J PEDIATR 1990; 117:261-3), came to the conclusion that the lactose content of human milk was not the culprit in colicky babies. This is contrary to our experience. From March 1987 to April 1988, eleven infants with colic participated in a double-blind, prospective, crossover study in which lactase (i n the form of LactAid) or placebo was orally administered for 5 days each. The laetase was given before the initiation of breast-feeding and again in mid feeding. One child was lost to follow-up; seven of the remaining 10 infants responded to the active enzyme. Four of those seven had such a marked and reproducible response that the mothers insisted on additional treatment after the 10-daY study. Although we were not able to do breath hydrogen tests, three of the infants who had a small amount of reducing substance in the stool before the test had a clinical response to the lactase. Our data support the concept that in some breast-fed babies, lactose intolerance is the cause of colicky behavior. The manner in which the lactase was administered--that is, before and again during breast-feeding--seems to be the only difference between our study and Dr. Miller's study; he gave the enzyme at the start of a feeding only. Although Dr Miller's dosage was based on an in vitro test demonstrating 75% hydrolyzation by 25 minutes, the dosage might have been less than needed, as demonstrated by the lack of decrease in the breath hydrogen levels. Harry F. Laws II, Colonel, USAF, M C Pediatrician, 554th Medical Group Nellis AFB, N V 89191-5300
Reply To the Editor: Dr. Laws suggests that lactase may be effective if administered before and during breast-feeding but not when the enzyme is given within 5 minutes of the start of a feeding. It is possible that giving yeast lactase, which is unstable at acid pH, within 5 minutes of the start of a feeding results in inactivation of the enzyme in the stomach. However, a recent study by Moore, 1 involving 39 breast-fed infants who were given lactase toward the end of each feeding, also failed to demonstrate an effect of lactase on the duration of crying and fussing or breath H2 levels. Eight formula-fed infants were also studied. These infants were fed either a lactose-hydrolyzed formula (LHF) or a lactose-containing formula (LCF) for 1 week and then changed to the other formula for a further week. There was no significant difference in the duration of crying and fussing between the babies who who were given LHF and those given LCF. No reduction in breath H2 excretion was observed in the babies receiving LHF. Although reducing lactose intake may not be an effective treatment for colic, it may be prophylactic. We followed 28 healthy infants who were fed either an LHF or an LCF from the first week
