THE CIRCUMCISION REFERENCE LIBRARY
SIR,—Your May 12 editorial about case-control studies and the creativity of "hundreds of small research studies" prompts us to record an example of what may be further evidence of the potential of the case-control approach. Discussion with medical students, engaged in basic epidemiology training, of a Lancet paper on breastfeeding and bacterial adhesion to uroepithelial cells1 led to a case-control study of exposure to breastfeeding and urinary tract infection (UTI) in infants. It is not known if human milk protects against UTI 2.
The 62 cases were all infants aged 0-6 months who were admitted to the university department of paediatrics between Jan 1, 1980, and Dec 31, 1988, and whose diagnosis was UTI, confirmed by at least one urine culture, done in our hospital, that yielded more than 100000 organisms/ml of a single species. The children had one or more of the following: failure to thrive, vomiting, diarrhoea, and fever. Two groups of controls -- matched by sex, age (within 15 days), father's occupation, geographical area of residence, birth order, type of delivery, and month of admission -- were chosen by systematic sampling. 62 controls were seen at our well-baby clinic; they were healthy and no urine cultures were done (in healthy infants asymptomatic bacteriuria is very unusual3). The other 62 controls had been admitted to our hospital with an acute illness and had negative urine cultures. None of the males, cases or controls, had been circumcised.
Data on breastfeeding were complete from the clinical records of both cases and controls. The proportion of infants who had been put to the breast at all was 47% in the cases and 82% and 87% in the controls (p < 0.001), and the mean duration of breastfeeding was 60, 61 and 61 days. For the purpose of analysis we defined an infant as breast-fed if he or she had had breast milk up to the day of the admission or outpatient visit (or at least 3 days before). The factors for which we matched cases and controls are not confounders for UTI 4 so the table gives the results for the unmatched analysis; matched and unmatched relative risk estimates were almost identical.
CHARACTERISTICS OF CASES AND CONTROLS __________________________________________________________________________ | | Well-baby | Inpatient Characteristic | Cases | controls | controls ------------------------------|----------|----------------|--------------- Age | | | 0-1 mo | 9 | 9 | 9 2-3 mo | 30 | 30 | 30 4-6 mo | 23 | 23 | 23 Sex (M/F) | 36/26 | 31/31 | 35/27 Social Class: | | | I-III | 22 | 22 | 20 IV-VI | 36 | 37 | 39 Unknown | 4 | 3 | 3 Birth Order | | | (first/subsequent) | 26/36 | 26/36 | 26/36 Breastfeeding (see text) | 10 (16%) | 34 (55%) | 26 (42%) ------------------------------|----------|----------------|--------------- Odds Ratio (and 95% confidence interval) | 6.3 (2.5-16.1) | 3.7 (1.5-9.5) -------------------------------------------------------------------------- |
These are the first data suggesting a protective effect of breastfeeding against UTI. Two medical students in two weeks came up with a clinically and biologically interesting hypothesis: will it be a challenge for "burnt-out" investigators who, you say, are fit only for the administration of confirmatory randomised trials?
Alfredo Pisacane
Liberatore Graziano
Gregorio Zona
Department of Paediatrics,
University of Naples,
80131 Naples, Italy
[CIRP Note: Pisacane et al. later published a more detailed and conclusive study. See Breast-feeding and urinary tract infection]
http://www.cirp.org/library/disease/UTI/pisacane/