Evaluation of Data Completeness in the Maryland Violent Death Reporting System: Variability Among Demographic Groups

BACKGROUND: The Maryland Violent Death Reporting System (MVDRS), which compiles detailed information for all in-state deaths of Maryland residents categorized as homicides, suicides, accidental firearm deaths, or deaths of undetermined intent, is an effective tool for violence research and prevention. As such, complete and accurate information is vital. This study evaluated two attributes – data quality (overall completeness of case records) and representativeness (completeness across demographic groups) – as set forth by the Centers for Disease Control and Prevention’s (CDC’s) Updated Guidelines for Evaluating Public Health Surveillance Systems. 

METHODS: All deaths recorded in MVDRS for the complete years 2003-2011 were analyzed. Record completeness was defined as having “precipitating circumstances” data, gleaned from the medical examiner report, police report, or both, if available. Circumstances were coded as “known” if at least one circumstance was entered, even if one or more additional details were unknown. Frequency of having known circumstances was assessed overall, across manners of death, and across demographic groups (race/ethnicity, gender, age category, and county of residence). Circumstance data for homicide was further investigated with logistic regression, using all demographic factors and one situational factor (physical location where incident occurred) as predictors. 

RESULTS: For all non-homicide manners of death (4,351 suicides, 20 accidental firearm deaths, and 5,743 deaths of undetermined intent), ≥85% of records contained circumstances data. However, only 50.5% of the 4,211 total homicide records contained this information. In bivariable analyses, males, Blacks, and Latinos were less likely than their counterparts to have circumstance data, regardless of manner of death, and there were significant differences by age group and county. In multivariable analyses for homicide specifically, Black race, younger age groups, living in a county with a high homicide rate, and dying outside in a public area were significantly associated with having circumstance data, when controlling for the other factors (p<.0001 for each). 

CONCLUSIONS: Data were significantly more complete for non-homicide violent deaths compared to homicide deaths. Demographic differences in completeness existed and were most pronounced for homicide. Disparities were not due to geography alone; Blacks and younger victims were less likely to have any circumstance data for homicide, even when controlling for residence in a county with a high homicide rate. Notably, the groups least likely to have complete records are those facing the highest burden of homicide. More complete data for the most affected groups should be prioritized in order to allow for the design of targeted interventions.