• L H Harrison, S Moursi, A H Guinena, A M Gadomski, K S el-Ansary, N Khallaf, and R E Black.
• Department of International Health, Johns Hopkins University School of Hygiene and Public Health, Baltimore, MD 21205, USA.
• Int J Epidemiol. 1995 Oct 1; 24 (5): 1058-63.

BackgroundAcute respiratory infection (ARI) is a major cause of childhood morbidity and mortality in developing countries. Community surveys are used to determine the proportion of children with ARI for whom care is sought by questioning mothers about the signs and symptoms of illness episodes. The validity of this approach has been studied infrequently.MethodsWe evaluated maternal reporting of signs and symptoms 2 and 4 weeks after diagnosis among 271 Egyptian children < 5 years old. Children with ARI were evaluated by physical examination, chest radiography, and pulse oximetry, and were alternately assigned for a maternal interview about the episode 14 or 28 days later.ResultsFor radiographically-defined acute lower respiratory infection (ALRI), the sensitivity of several symptoms for combined open- and close-ended questions was relatively high: nahagan (deep or rapid breathing) (80%), nafas sarie (fast breathing) (66%), and kharfasha (coarse breath sounds) (63%). The specificity of these terms was 50-68%. The specificity was inversely related to the follow-up time. No term provided both a sensitivity and specificity of > 50% at day 28 across the radiographically, clinically- and pulse oximetry-based definitions of ALRI. Spontaneously mentioned karshet nafas (difficult or rapid breathing) at 14 days had a specificity and sensitivity for radiographic ALRI of 87% and 41%, respectively, suggesting that this term is a good choice for community surveys.ConclusionsMaternal reporting of ARI symptoms is non-specific 2 and 4 weeks after diagnosis but may be useful for monitoring trends in the proportion of children with pneumonia who receive medical care. To maximize specificity, ARI programmes should generally use a recall period of 2 weeks.

51 keywords...

hide…