Intended for healthcare professionals


Breast feeding and multiple sclerosis

BMJ 1994; 308 doi: (Published 28 May 1994) Cite this as: BMJ 1994;308:1411
  1. A Pisacane,
  2. N Impagliazzo,
  3. M Russon,
  4. R Valiani,
  5. A Mandarini,
  6. C Florio,
  7. P Vivo
  1. Departmento di Pediatria Universitia di Napoli Federico II, Via Pansini 5, 80131 Naples, Italy.
  2. Divisione di Neurologia Ospedale Antonio Cardarelli, Napoli e Associazione Italiana Sclerosi Multipla, Naples, Italy.
  • Accepted 17 December 1993

Several studies indicate that multiple sclerosis might be a rare result of delayed exposure to a common infectious agent,1 but some authors have suggested a “geographical predisposing factor” that could be associated with the diet in industrialised countries.2 We investigated the association between multiple sclerosis and breast feeding in a case-control study.

Patients, methods, and results

Cases and controls were selected from the patients attending the neurological outpatient department of Cardarelli General Hospital, Naples, where neurologists from the Italian association for multiple sclerosis follow up about 300 patients with this disease. Cases were all the patients observed between 1 March and 31 May 1993 who had a definite diagnosis of multiple sclerosis according to the criteria of Poser et al.3 Controls were the first patients observed in the same department during the same period who matched the cases in terms of age (plus or minus five years) and sex. Patients who declined to be interviewed or could not provide reliable information about their feeding during infancy were excluded.

The groups were contacted by two interviewers, who were blind to the objectives of the study and interviewed the same number of patients from each group. Whenever possible the mothers of the patients were interviewed, otherwise the patients provided the information, usually quoting their mothers. Relative risk was calculated by odds ratios with confidence intervals by Cornfield's method. Confounding and effect modification were investigated by stratified analysis. Ninety three cases and 93 controls were enrolled in the study. The diagnoses of controls were back pain and sciatic nerve injury (56), hemicrania (19), polyneuropathies (15), and other (four). The information was provided by the mothers for 56 (60%) cases and 54 (58%) controls.

The table shows characteristics of the groups. The mean (SD) duration of breast feeding was 8.4 (6.9) months for cases and 12.5 (7.5) months for controls (t=-3.9, df=184, P=0.001). The distribution of feeding was the same whether data were provided by the mothers or by the patients. Stratified analysis showed that no factor among those we analysed (birth weight, type of delivery, social class, sex, age group, number of other children in the household) confounded or modified the association between feeding and illness.

Characteristics of patients with multiple sclerosis according to case-control status

View this table:


Our data indicate that patients with multiple sclerosis were less likely than controls to have been breast fed for a prolonged period of time. Selection bias is unlikely because cases and controls were contacted in the same department and their mothers were all contacted by telephone by the same interviewers. The validity of information on feeding may be a limit of our methods, but there is no reason to suspect a differential recall between cases and controls.

There are several reasons why prolonged breast feeding may be associated with a decreased risk of multiple sclerosis. Cow's milk contains lower amounts of unsaturated fatty acids, and a different composition of cortex grey matter has been described in bottle fed infants.4 This fact could be associated by means of the formation of defective membranes with easier entry of an infective agent across the blood-brain barrier or with accelerated degradation of myelin itself.2 Human milk might actively influence the immune system of the offspring by different mechanisms,5 and some features of the immune response among those who have been breast fed for a prolonged period may last for a long time. Alternatively, prolonged breast feeding may be a marker of some unknown socioeconomic characteristic that could be associated with a low risk of illness. The prevalence of prolonged breast feeding in industrialised countries is currently much lower compared with the time when our patients were born. Should our data be confirmed another point would be added to the long list of the benefits of prolonged breast feeding.


View Abstract