Agreement between clinical examination and parental morbidity histories for children in Nepal

Parental histories are often used to estimate the prevalence and the impact of interventions on child morbidity, but few studies have examined the agreement between parental histories and clinical examination. We compared clinical findings with a same-day parental morbidity history for preschool-age children in rural Nepal. A 15 per cent sample of children from 40 wards in Sarlahi district, Nepal, was selected for participation and 814 same day morbidity histories were obtained from parents. A clinician, masked to the parent's history, visited the household 2-4 h later and examined the child for signs of morbidity symptoms about which the parent had previously been questioned. Signs included measurement of temperature, respiratory rate, examination of stools, ear discharge, and presence of persistent cough. Agreement between the history and clinical examination was excellent for ear infection (κ = 0.75) and history of measles rash (κ = 0.74), moderate to poor for diarrhoea (κ = 0.21) and fever (κ = 0.31), and there was no evidence of agreement for dysentery (κ = -0.01), rapid breathing (κ = 0.06), and cough (κ = 0.09). The prevalence of dysentery, fever, cough, and rapid breathing was lower if clinical signs rather than histories were used. The prevalence of diarrhoea was higher if the presence of a loose stool in a cup rather than a history was used. The prevalence of ear infections and measles was comparable with both methods. The agreement between histories and clinical examination varies by morbidity type, as does the prevalence of morbidity estimated by one or other method.