JOURNAL OF PEDIATRICS, Volume 120, Number 1: Pages 87-89,
January 1992.

Breast-feeding and urinary tract infection

Alfredo Pisacane, MD, MSc, Liberatore Graziano, MD
Gianfranco Mazzarella, MD, Benedetto Scarpellino, MD, and
Gregorio Zona*

From the Dipartimento di Pediatria, Università di Napoli, Napoli, Italy.


A case-control study was conducted to study the association between breast-feeding and urinary tract infection. Case patients were 128 infants aged birth to 6 months with urinary tract infection. Control infants were 128 infants admitted to the same ward with an acute illness. The results support the hypothesis that breast-feeding protects infants against urinary tract infection. (J PEDIATR 1992;120:87-9)

Much evidence from developing countries shows that breast-feeding protects infants from infection,1 but evidence from developed countries is less convincing.2 Recently Bauchner et al.3 reviewed a selection of 20 studies on this subject written in English since 1970; they concluded that most of the studies had major methodologic flaws and that breast-feeding had a minimal protective effect in industrialized countries. Recent studies which close attention was paid to the methodologic criteria of Bauchner showed contrasting results.4,5

UTI Urinary tract infection

No data are available about the effect of breast-feeding in protecting against urinary tract infection; however, some authors reported the presence of an oligosaccharide in the urine of breast-fed infants that caused inhibition of Escherichia coli adhesion to uroepithelial cells.6 Other researchers showed an increased concentration of immunologic factors in the urine of breastfed infants.7,8 In this report we present the results of a case-control study of the relationship between UTI in the first 6 months of life and infant feeding.

Table. Characteristics of case and control groups ----------------------------------------------------------------------- Characteristic Case Control (n=128) (n=128) ----------------------------------------------------------------------- Age (mo) 0-1 21 21 2-3 50 50 4-6 57 57 Gender (M/F) 81/47 81/47 Social class I-III 39 38 IV-VI 80 79 Unknown 9 11 Birth order (first/subsequent) 48/80 53/75 Maternal smoking status Nonsmoking 69 75 Smoking <15 cigarettes/day 25 17 Smoking >15 cigarettes/day 15 13 Unknown 19 23


Study area. The Medical School of Naples Department of Pediatrics admits to the hospital approximately 2000 children per year; 15,000 children are seen every year as outpatients. More than 95% of these children come from Campania, one of the largest and most populated regions of southern Italy (13,600 km2, with 5.5 million inhabitants).

Selection of case and control patients. Case patients included all 128 hospitalized infants aged 6 months and less in whom UTI was diagnosed from the beginning of January 1976 to the end of December 1989. We ascertained UTI by two urine cultures done at our hospital that yielded more than 100,000 colonies/ml of a single bacterial species. According to the policy of our department, all urine specimens were collected by trained nurses. After cleansing the infant's perineal area with castile soap and water, a pediatric nurse applied a sterile pediatric urine collector. The bag was removed no more than 30 minutes after application and was taken to the hospital laboratory.

Five infants with low birth weight and three for whom no detailed information was present in the clinical chart were not included in the study. One control patient for each case patient was chosen by systematic sampling from among infants who had been admitted to our hospital with an acute illness between Jan. 1, 1976, and December 31, 1989, and who had one urine culture, done with the same technique as described for the case patients, with negative results. The control patients were matched to the case patients by gender, age (within 15 days), and year and month of admission; the next infant who met the matching criteria was selected. Infants with diagnoses of acute diarrhea and respiratory infection were not included as control patients because of evidence that breast-feeding confers some protection against those diseases. All control patients had a birth weight >2500 gm.

Classification of feeding methods. The clinical charts used for children admitted to hospital contain a sheet with detailed feeding information; almost 100% of our records give accurate information about feeding during infancy.

The categories of feeding were classified as follows: (1) 100% breast-feeding; (2) breast-feeding combined with bottle feeding (ration not available from the records); and (3) 100% bottle feeding.

For both case and control patients, type of feeding was defined as the type that the infant was receiving at the time of hospitalization. In addition, a dichotomous classification of infants as either breast-fed at least somewhat or never breast-fed was done to minimize reverse causality bias. No information about feeding was available for two hospitalized control patients; they were removed from the study and replaced by the two next babies who met the matching criteria and for whom feeding information was available.

Data Collection and analysis. For all subjects, data concerning background variables (age, gender, birth weight, social class, birth order, maternal smoking habits), disease status (outcome), and feeding method. (exposure) were ascertained from the hospital charts. After the case and control groups had been selected, the sheets containing the information about feeding were duplicated and analysed by two of us who were unaware of the disease status. In addition, 10 randomly selected families in each group were contacted at home by a pediatric nurse; information about feeding during the first six months was collected from them. These data were concordant with those in the clinical charts for all children studied.

Relative risk was estimated by the odds ratio; 95% confidence intervals were calculated with a test-based method.9 Because the odds ratios did not differ significantly for the two categories of breast-feeding plus bottle feeding), full and partial breast-feeding groups were combined into a single group for comparison with the bottle-fed group.


The characteristics of the study groups are shown in the Table. All male infants were uncircumcised. The main causes of admission among the case patients were as follows: failure to thrive, vomiting, diarrhea, loss of appetite, and fever. The main admission diagnoses among control patients were as follows: failure to thrive, anemia, cardiac diseases, and congenital malformations.

In the case group 73% of the urine cultures yielded Escherichia coli, 17% grew Proteus species, and 10% grew other bacterial species. The numbers of infants who had ever been breast fed were 64 of the case patients (73%) (p<0.001). Sixteen case patients (12%) and 56 hospitalized control patients (44%) were being breast fed at the time of observation (p<0.001). Compared with infants who were bottle fed, breast-fed patients had a relative risk of UTI of 0.38 (95% confidence intervals 0.22 to 0.65) when a dichotomous classification (ever breast fed vs never breast fed) was used. The odds ratio relative to breast-feeding status of the time of admission was 0.18 (95% confidence intervals 0.09 to 0.36).


The results of our study suggest that breast-feeding provides substantial protection against UTI during the first 6 months of life among children in an industrialized country, confirming our preliminary report.10 Indirect evidence of such protection has been offered by others,6-8 but only one clinical study11 is available and it had many flaws, caused by unclear definitions of exposure and disease and its small study groups. The roles of bias and confounding in the studies on breast-feeding and infections have been widely discussed; several cohort and case-control studies have been judged inadequate because of detection biases; unclear definition of exposure, outcome or both; and lack of adjustment for potential confounding variables.3

We attempted to improve methodologically on past studies according to the indications of Bauchner et al.3 Detection bias is not likely to have occurred in our study because our policy is to hospitalize only severely ill infants, with the decision made independently of the feeding regimen. The hospital control group has a proportion of breast-fed infants similar to that of the population of infants in our well-baby clinic (unpublished data). Protopathic (reverse causality) bias may stem from the fact that an infant's health can affect the choice of a feeding method. The incidence of breast-feeding among our case patients was significantly lower than that among control patients. This fact, rather than the interruption of breast-feeding because of disease, seems to be the main reason for the difference in breast-feeding prevalence observed at the time of admission to the hospital.

Confounding variables for UTI have not been described12; our case and control groups were similar with regard to the variables described as confounding for the association between breast-feeding and infections3; even if unknown confounding variables are postulated for this association, it is unlikely that they account for the size of the effect observed.13

We are aware that a suprapubic tap is more reliable than bag collection for the diagnosis of UTI.14 Nevertheless, the specimens from case and control patients were collected by the same persons, with the same technique and at the same time; all case patients had symptoms and required hospitalization. Recent data show that the external bacterial contamination rate for urine collected by the bag technique from non-toilet trained infants is not significant.15

We did not investigate whether the protection conferred by breast-feeding persists after breast-feeding has been interrupted, although preliminary data seem to suggest this inference.16 We conclude that breast-feeding seems to protect against UTI during the first 6 months of life.


  1. Jason JM, Nieburg P, Marks JS. Mortality and infectious disease associated with infant-feeding practices in developing countries. Pediatrics 1984;74:702-27. [Abstract]]
  2. Kovar MG, Serdula MG, Marks JS, Fraser DW. Review of the epidemiologic evidence for and association between infant feeding and infant health. Pediatrics 1984;74:615-38. [Abstract]
  3. Bauchner H, Leventhal JM, Shapiro E. Studies of breastfeeding and infections: How good is the evidence? JAMA 1986;256:887-92. [Abstract]
  4. Howie PW, Forsyth JS, Ogston SA, Clark a, du V Florey C. Protective effect of breast feeding against infection. BMJ 1990; 300:11-6.
  5. Rubin DH, Leventhal JM, Relationship between infant feeding and infectious illness: a prospective study of infants during the first year of life. Pediatrics 1990; 85:464-71.
  6. Coppa GW, Gabrielli O, Giorgi P, et al. Preliminary study of breastfeeding and bacterial adhesion to uroepithelial cells. Lancet 1990; 335:569-71.
  7. Goldblumm MR, Schanler RJ, Garza C, Goldman AS. Human milk enhances the urinary excretion of immunologic factors in low birth weight infants. Pediatr Res 1989;25:185-9.
  8. Prentice A. Breast feeding increases concentrations of IgA in infant's urine. Arch Dis Child 1987;62:792-5. [Abstract]
  9. Meittinen OS. Estimability and estimation in case-referent studies. Am J Epidemiol 1976; 103:226-35.
  10. Pisacane A, Graziano L, Zona G. Breast-feeding and urinary tract infection [Letter]. Lancet 1990; 336:50.
  11. Cheong CIT. Human breast milk in Escherichia coli urinary tract infection. Milit Med 1982;147:202-4.
  12. Kunin CM. Epidemiology and natural history of urinary tract infection in school age children. Pediatr Clinic North Am 1971;18:509-16.
  13. Angell M. The interpretation of epidemiologic studies. N Engl J Med 1990;323:823-5.
  14. Kunin CM. Urinary tract infections in infancy. J Pediatr 1975;86:483-4.
  15. Schlager TA, Dunn ML, Dudley SM, Lohr JA. Bacterial contamination rate of urine collected in a urine bag from healthy non-toilet-trained male infants J Pediatr 1990; 116:738-39.
  16. Marild S, Jodal U, Hanson LA. Breast-feeding and urinary tract infection [Letter]. Lancet 1990;336:942.

Submitted for publication May 29, 1991; accepted July 15, 1991.

Reprint requests: Alfredo Pisacane, MD, MSc, Dipartimento di
Pediatria, Universita di Napoli, Via Pansani 5, 80131 Napoli, Italy.

*Medical Student

(File revised 15 September 2006)

Return to CIRP library