Overall, 1,098 cases of asthma were identified by the mother, the medical provider, or both. For 36% of those cases, the mother and medical provider were in agreement: both said that the child had asthma. For the remaining 64% of cases, however, the mother and doctor disagreed, with one but not the other source reporting asthma.

Cohen’s kappa statistic for concurrence between maternal reports and medical records was 0.48, indicating moderate agreement according to standard criteria described by Landis and Koch. Cohen’s kappa is based on the difference between the observed proportion of subjects upon whom the two reporting sources agree, and the proportion expected by chance, given the marginal distributions (e.g., the distribution within each source alone).

Related readings:

Landis JR, Koch GG, 1977. “The measurement of observed agreement for categorical data,” Biometrics. 33:159-74.