The recent study by Paunio et al. ( 1) reported results based on a cross-sectional immunization study from Finland that included data on parentally recalled lifetime incidence of doctor-diagnosed measles and atopic diseases. The authors observed a positive association between measles and the risk of asthma, atopic eczema, and allergic rhinitis. Although they ack-nowledged that the results do not prove a causal relationship, they questioned the hypothesis that measles infection is protective against atopy ( 2). Fur-thermore, they questioned the validity of the “hygiene theory” behind the increase of asthma and atopy ( 3).

We propose that differential misclassification of exposure ( 4) may explain the positive association found in that study ( 1). There are two possible ways in which this may occur:

  • underdiagnosis of measles especially among healthy, nonatopic children

  • overdiagnosis of measles among children with asthma or atopy.

Paunio et al. ( 1) suggest that selection bias due to a higher measles mortality rate among atopic children may explain the inverse association between measles and atopy found in Africa ( 2). This argument is based on findings that atopic children may be more vulnerable to measles infection ( 5), perhaps due to lower production of interferon-gamma and decreased virus clearance. By the latter argument, atopic children have more serious illness due to measles infection, seeking medical care more often than their nonatopic counterparts. This would lead to diagnostic bias in favor of atopics. In other words, if nonatopics more often had mild or subclinical measles, the probability of the measles diagnostic label would be obviously lower, as the children may have been cared for at home and never taken to the physician. The work of Hayney et al. ( 5) suggests that this option is plausible and likely.

Option 2) given above requires that atopic children be more often in contact with a physician (with higher probability of being diagnosed with measles), but also that they have other clinical signs (such as skin reactions due to other viruses) that were mistakenly diagnosed as measles. This option requires rather large, differential overdiagnosis of measles among atopic children, an occurrence which we do not assume to be a feasible explanation.

In conclusion, we propose that differential misclassification of exposure, namely, the higher level of underdiagnosis of measles among nonatopic children, cannot be ruled out as a potential noncausal explanation for the results of Paunio et al.


  1. Top of page
  2. References
  • 1
    Paunio M, Heinonen OP, Virtanen M, Leinikki P, Patja A, Peltola H. Measles history and atopic diseases. A population-based cross-sectional study. JAMA 2000;283:343 346.
  • 2
    Shaheen SO, Aaby P, Hall AJ, et al. Measles and atopy in Guinea-Bissau. Lancet 1996;347:1792 1796.
  • 3
    Martinez FD & Holt P. Role of microbial burden in aetiology of allergy and asthma. Lancet 1999;354 Suppl 2:12 15.
  • 4
    Rothman KJ & Greenland S. Precision and validity in epidemiologic studies. In: RothmanKJ, GreenlandS. Modern epidemiology. 2nd ed. Philadelphia, PA: Lippincott-Raven, 1998:115 134.
  • 5
    Hayney MS, Poland GA, Jacobson RM, et al. Relationship of HLA-DQA1 alleles and humoral antibody following measles vaccination. Int J Infect Dis 1998;2:143 146.