- Alfredo Pisacane, senior lecturera,
- Ugo de Luca, senior surgical registrarb,
- Luciano Criscuolo, postgraduate traineea,
- Filomena Vaccaro, postgraduate traineea,
- Adriana Valiante, postgraduate traineea,
- Angelina Inglese,
- Giuseppe Caracciolo, consultant surgeonb,
- Luciano Pinto, consultant neonatologistb
- a Department of Paediatrics, University of Naples, 80131 Naples, Italy
- b Santobono Paediatric Hospital, Naples
- Correspondence to: Dr Pisacane.
- Accepted 23 November 1995
Hypertrophic pyloric stenosis is a common infantile disorder characterised by enlarged pyloric musculature and gastric outlet obstruction. Although there is a genetic component,1 some evidence suggests that postnatal factors may also play an important part. Pyloric stenosis has been associated with birth rank, maternal anxiety and drug use, and feeding practices.2 We investigated the association between pyloric stenosis and infant feeding in a population based casecontrol study.
Patients, methods, and results
All 102 infants admitted to Santobono Paediatric Hospital, Naples between 1 July 1992 and 31 December 1993 with surgically confirmed pyloric stenosis were recruited for the study. All were born and lived in Campania province. Controls were 204 infants randomly selected from around 600 enrolled from a provincial birth list for a child health survey; all were born in 1993 and were matched to cases of the same sex. Two medical students, unaware of the objectives of the study, interviewed the mothers at around the first birthday of their infants. Sample size was based on the assumption that a 15% difference in the prevalence of breast feeding between the two groups would be clinically significant. Relative risk was estimated by the odds ratio; 95% confidence intervals were calculated by Cornfield's method. The potentially confounding variables of sex; birth order; birth weight; maternal education, age, and smoking; and type of birth were studied by stratified analysis. The percentage population attributable risk was calculated on the assumption that the controls' feeding pattern was representative of that of the general infant population of our province.
Eighty three cases were boys, giving a male:female ratio of 4.4:1. Two thirds of the infants started vomiting during the first four weeks of life but only seven (6.8%) during the first week. Several characteristics (birth weight; birth order; type of birth; maternal age, education, and smoking during pregnancy; family history of allergic disorders; and infant allergic disease until the first birthday) were similar in cases and controls. Table 1 shows the relative risk of pyloric stenosis analysed by the type of milk consumed. The association between breast feeding and disease was neither confounded nor modified by the factors we analysed. The proportions of controls who were bottle fed (12.2%; 25/204) or partially breast fed (20.6%; 42/204) at 1 week of age gave a percentage population attributable risk of 35% (95% confidence interval 6% to 72%).
This study indicates that infants with pyloric stenosis were less likely to have been exclusively breast fed during the first week of life. Our data are not consistent with those of Dodge3 and Webb,4 who reported that breast feeding was more common among such infants. The design of our study probably avoided selection bias as well as recall bias. Moreover, because in most patients the symptoms were not yet present at the age of 1 week, reverse causality bias was also unlikely.
Explanations for a protective effect of breast feeding in pyloric stenosis can only be speculative but include the presence in human milk of high levels of hormones like vasoactive intestinal peptide that can favour pyloric relaxation; the increased plasma gastrin concentrations in infants fed by bottle, wholly or partially,5 which might be associated with pylorospasm, pyloric hypertrophy, and consequent damage of peptide containing nerve fibres; and, finally, the presence of unknown confounders. If this association is causal up to 35% of cases of pyloric stenosis in our population might be attributed to the lack of exclusive breast feeding during the early neonatal period. In view of conflicting results from other studies, however, and in the absence of biological supporting data, a causal interpretation for this association is not yet warranted.
Funding Department of Paediatrics, University of Naples.
Conflict of interest None.