Discussion

We observed that the most frequent complaints were fever and respiratory signs (cough or difficult breathing and runny nose) in both age groups and that approximately half of all children had more than one complaint. In addition, diagnoses related to respiratory problems, either primary or secondary, were the most frequent ones. Sensitivity of mother’s complaint to detect the presence of fever was moderate and specificity was low. Triage seemed not to influence children’s outcome.

Few young infants (< 2 mo) were enrolled. They accounted for only 5% of all visits. This age group is of high risk for mortality. The young infant may be less frequently ill or may be less frequently brought to be seen when ill. In other settings young infants seem to be more frequently seen in primary care facilities(4) although this needs further investigation. In all situations, however, an important question is to know where mothers seek care when their young infant is ill. Further studies will need to address this issue.

Most frequent complaints correlated well with pediatrician’s diagnosis and were similar to the sequence proposed in the IMCI algorithm. In addition, the frequency of more than one complaint support the MCI rationale. It is worth to note that accidents, that are not part of the IMCI algorithm, were an important complaint and diagnosis in our setting. However, most accidents were not severe. In addition, severe malnutrition seems not to be an important problem, as it is in other settings(5), although nutritional status was not assessed and some severe cases could have been overlooked. These findings suggest that large urban areas, usually more developed and more populated than rural ones, have a different profile of clinical syndromes. This indicates that, to be effective, adaptations are needed in the IMCI algorithm to take care of this profile.

Respiratory syndromes were still the main cause of consultation and complaint, even considering that data were collected during the hot season in Rio. These findings were similar to what was observed in less populated and poorest areas(6,7). In the young infant, wheezing was not present, contrary to what was observed in children above 2 months of age. Wheezing was more frequently seen than pneumonia, although pneumonia was the third most frequent respiratory tract diagnosis among children above 2 months of age. Wheezing seems to be highly prevalent in children in Rio de Janeiro(8). These findings show the importance of determining the correct diagnosis in cases of fast breathing given the need for a correct and appropriate treatment. Late detection of pneumonia cases would increase the risk of death. These results suggest that wheezing should be included in the IMCI algorithm when used in Rio de Janeiro. In fact, the adapted IMCI Brazilian algorithm includes the assessment and treatment of wheezing in children older than 2 months(9).

When mothers complained that the children had fever, they rarely missed a child with fever, but they overestimated the number who had fever. This suggests that fever complaint, as it was measured, may not be a good triage sign because it would overburden the health services because of the high number of false positive cases. However, we calculated sensitivity taking the presence of fever (> 37.5 degrees) at the time of consultation as gold standard. This could have lowered sensitivity, as many patients      who are reported to have fever might not be having fever at the time of consultation. Other authors studied mother’s assessment of fever during consultation and found higher sensitity(10,11).

We did not observe any difference among discharged children in relation to the presence of triage. We would expect that triaged sick children would be more severe and thus would have a worse outcome. However, the sample size was probably small to detect such a difference. In addition, our sample was selected on a convenience basis. This could have accounted in part for the observed results, in special in the subset of young children. Although this is a potential bias, we do not believe it would change the general trend of our findings.

Another limitation may be related to the characteristics of our Emergency Unit. As it is part of a university tertiary hospital this profile of complaints and diagnosis may not reflect what is occurring in the secondary level or in general hospital in Rio. Although other studies are needed to clarify this issue we believe the general trend we observed would not be changed, as most patients are brought by their mother to our Emergency Unit spontaneously, i.e., they are not referred.

In conclusion, our findings support the proposed WHO-IMCI strategy to decrease mortality in childhood. However, some adapta-tions are needed when implementing the use of the algorithm in Rio de Janeiro and possibly in other similar large urban areas.

Contributors: AJLAC designed, coordinated the study, analyzed the data and drafted the paper; he will act as the guarantor for the paper. RA, CSO, AMGS and EG participated in date collection, and also helped drafting the paper. RA helped supervising data collection and data entry.

Funding: Universidade Federal do Rio de Janeiro.
Competing interests: None stated.