OBJECTIVE:To analyze the coverage and quality of the data on hospitalizations due to external causes in the Hospital Information System of the Brazilian National Health System. METHODS:Hospitalizations recorded in the Hospital Information System of 11 hospitals in two municipalities (Londrina and Maringá, Southern Brazil), in 2004, were compared with hospitalizations identifi ed as consequences of external causes through investigating medical records, complemented with mortality and pre-hospital data. The crude agreement rate, sensitivity and positive predictive value were calculated. The profi le of groups of hospitalization causes recorded in the System was compared with that obtained from investigations. RESULTS:In Londrina, 3,002 hospitalizations due to external causes were recorded in the System and, in Maringá, 1,403. The investigations found 4,018 and 2,370, respectively. The System presented high positive predictive values for hospitalizations due to external causes in both municipalities: 97.7% in Londrina and 98.6% in Maringá. However, the sensitivity was low: 57.3% in Maringá and 73% in Londrina, thus denoting underrecording of these causes. Comparison between the profi les of the types of causes revealed that there was underestimation of some causes in the System, especially regarding accidents due to exposure to inanimate mechanical forces, assaults and intentional self-harm. CONCLUSIONS:Underrecording of hospitalizations due to external causes and some distortions regarding the types of causes occurred in the Hospital Information System in both municipalities. Detection of these defi ciencies may contribute towards the process of improving the quality of the information in this System.

OBJECTIVE:To analyze the coverage and quality of the data on hospitalizations due to external causes in the Hospital Information System of the Brazilian National Health System. METHODS:Hospitalizations recorded in the Hospital Information System of 11 hospitals in two municipalities (Londrina and Maringá, Southern Brazil), in 2004, were compared with hospitalizations identifi ed as consequences of external causes through investigating medical records, complemented with mortality and pre-hospital data. The crude agreement rate, sensitivity and positive predictive value were calculated. The profi le of groups of hospitalization causes recorded in the System was compared with that obtained from investigations. RESULTS:In Londrina, 3,002 hospitalizations due to external causes were recorded in the System and, in Maringá, 1,403. The investigations found 4,018 and 2,370, respectively. The System presented high positive predictive values for hospitalizations due to external causes in both municipalities: 97.7% in Londrina and 98.6% in Maringá. However, the sensitivity was low: 57.3% in Maringá and 73% in Londrina, thus denoting underrecording of these causes. Comparison between the profi les of the types of causes revealed that there was underestimation of some causes in the System, especially regarding accidents due to exposure to inanimate mechanical forces, assaults and intentional self-harm. CONCLUSIONS:Underrecording of hospitalizations due to external causes and some distortions regarding the types of causes occurred in the Hospital Information System in both municipalities. Detection of these defi ciencies may contribute towards the process of improving the quality of the information in this System.