[Data comparison on homicide deaths between two information systems, Brazil]

Area of Science:

• Public Health
• Criminology
• Health Informatics

Background:

• Criticism exists regarding the quality of Brazilian criminal data, yet systematic studies are lacking.
• This study addresses this gap by comparing homicide data from two public information systems in Minas Gerais, Brazil.

Purpose of the Study:

• To systematically compare the completeness and accuracy of homicide data recorded by the Mortality Information System of the Ministry of Health (SIM/MS) and the Minas Gerais Military Police (PMMG).
• To evaluate how population size influences the discrepancies between these two homicide data collection systems.

Main Methods:

• Homicide death data from SIM/MS and PMMG in Minas Gerais (1979-1998) were analyzed.
• Comparisons involved absolute numbers, ratios based on municipal population size, and time-series linear regression models for homicide rates.

Main Results:

• In municipalities under 100,000 inhabitants, PMMG data showed higher completeness than SIM/MS.
• In larger municipalities (over 100,000 inhabitants), SIM/MS was more efficient but included external and undetermined violent deaths.
• PMMG's under-registration in larger cities may stem from delayed reporting of hospital deaths.

Conclusions:

• SIM/MS and PMMG represent lower and upper bounds for homicide deaths in larger cities, respectively.
• When both datasets are available, they should be analyzed comparatively.
• Users of single datasets must consider the identified system-specific biases and limitations.