Breast feeding and tonsillectomy

BMJ 1996; 312 doi: (Published 23 March 1996) Cite this as: BMJ 1996;312:746
  1. Alfredo Pisacane, senior lecturera,
  2. Nicola Impagliazzo, postgraduate traineea,
  3. Carmela De Caprio, postgraduate traineea,
  4. Luciano Criscuolo, postgraduate traineea,
  5. Angelina Inglese, postgraduate traineea,
  6. Inglese, postgraduate traineea

    Maria Claudia Mendes Pereira da Silva

  1. a Department of Paediatrics, University of Naples, 80131 Naples, Italy
  1. Correspondence to: Dr Pisacane.
  • Accepted 23 November 1995

Tonsillectomy is one of the commonest operations in children,1 performed mainly for recurrent tonsillar and ear infections and severe nasal airway obstruction. The pathogenesis of tonsillar disease is probably multifactorial, and interactions among the anatomy, microbiology, and immunology of these structures may be important.2 Infant feeding is one factor that may programme the immune response, and bottle feeding has been associated with diseases such as acute appendicitis.3 We performed a case-control study designed to investigate the relation between tonsillectomy and infant feeding.

Patients, methods, and results

The study was carried out in seven primary schools in Naples. The parents of all children were sent questionnaires asking whether the children had had their tonsils and adenoids removed, about maternal education and smoking, and the children's feeding habits. Cases were all children whose tonsils and adenoids had been removed before 31 December 1994. For each case three controls were randomly selected from the same class. The parents of cases received a second questionnaire about the operation. All data were collected between February and April 1995. Sample size was determined after a pilot survey to assess the prevalence of the operation and based on the assumption that a 10% difference in frequency of breast feeding between study groups would be clinically important. Relative risks were estimated by odds ratios with 95% confidence intervals. Stratified analysis and unconditional multiple logistic regression were used to investigate the role of confounding and interaction.

Of the 2999 questionnaires sent 2458 (82%) were completed. Overall 238 children (9.6%) had had their tonsils removed, the main reasons being recurrent throat and ear infection (95%) and airway obstruction (91%). The relation between breast feeding and tonsillectomy was stratified by each of the other risk factors without evidence of confounding. Logistic regression analysis showed that factors independently associated with tonsillectomy were: breast feeding, maternal smoking, higher maternal education, and female sex (see table 1). A dose response relation was observed between tonsillectomy and breast feeding.

Table 1

Characteristics of cases and controls. Values are numbers (percentages)

View this table:


Our data show that, after adjustment for several potential confounders, children whose tonsils had been removed were less likely to have been breast fed. The strengths of this study are the adequate sample size, the high response rate, the unbiased choice of controls, and the analysis of the potential confounders. The weakness is the long time between exposure and data collection, though there is no reason to suppose a differential recall between cases and controls. An additional problem is that the relation investigated is not between infant feeding and a disease but between feeding and a surgical intervention. The operation is a proxy for recurrent respiratory infection but is also influenced by other socioeconomic and medical factors, such as operating time, bed space, and patient demand.4 The characteristics of the families who take the step from recurrent respiratory infection to tonsillectomy represent unknown potential biases.

The protection conferred by human milk against respiratory infection may persist for several years.5 The immune components of human milk provide an antigen avoidance system that can decrease the severity of infection and probably the inflammatory reactions associated with it. This milder inflammatory response might programme the infant's immune system, its effects lasting for several years, and might be associated with more tolerant tonsillar lymphoid tissue. These findings raise intriguing questions about the potential effects of different components of breast milk on the development of tissues and organs. Moreover, differences in the prevalence of breast feeding might help explain considerable regional variations in tonsillectomy rates.


  • Funding Department of paediatrics.

  • Conflict of interest None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract