The increased frequency and severity of atopic disease over the past few decades has been attributed by some to the “hygiene hypothesis,” which postulates that decreased exposure to common infections caused by vigilant hygiene causes an increased risk for development of atopy. It is known that viral and bacterial infections stimulate a subpopulation of T-lymphocytes that secrete cytokines that suppress production of IgE. Higher levels of IgE correlate with increased atopy.

Kilpi and colleagues analyzed atopy rates among 329 healthy infants from one town in Finland who were enrolled in a study following outcomes after uncomplicated pneumococcal infection. Each child was seen at least 10 times over the first 24 months of life, and whenever they were ill. At any sick visit, nasopharyngeal aspirate samples were taken for virus detection, and myringotomy fluid was sent for study whenever otitis media with effusion was seen. At each visit, the physical examination included an undressed assessment of the skin. Of the 329 infants enrolled, 85 percent completed the two years of follow-up.

A total of 837 viral infections and 871 episodes of acute otitis media were documented. Atopic dermatitis was diagnosed in 16 percent of study subjects, with the majority of cases presenting before six months of age. The rate of atopy was 39.4 cases per 100 infants per year when no infection was documented in the first six months of life, while those with an infection had rates ranging from 10.2 to 20.0 cases per 100 infant-years, depending on the type of infection (otitis media, viral infection, or respiratory infection). There was also a protective effect when the first infection occurred at more than six months of age, but the overall incidence of atopy was much lower in this older-infant group because most cases of atopic dermatitis had manifested by this time.

The investigators concluded that their results were consistent with the “hygiene hypothesis” that common, minor infections early in life may protect against atopic disease.

editor's note: Examining 95 percent confidence intervals for published study results is a chore that we are all tempted to gloss over, but it assumes greater importance when the groups being compared include smaller numbers of subjects. In this study, all of the 95 percent confidence intervals for comparisons between groups were wide, and they all included zero (i.e., they included the possibility that there was no protective effect).—b.z.