(Landen et al., 2003; Linakis and Frederick, 1993; Soslow and Wolf, 1992). Thus medical examiner data are not entirely captured in death certificates, while death certificate data include a far greater number of cases than those in medical examiner series.
Hospital discharge data systems (HDDS) at the local, regional, or state level represent a major source of health care data that potentially can be used for poisoning and drug overdose surveillance purposes (Agran et al., 2003; Hoyt et al., 1999; King, 1991; Smith et al., 1985, 1991; Sumner and Langley, 2000). The assignment of ICD-9 codes (routine and E-codes) draws on the direct medical evaluation documented in the hospital record. Hospital admission for poisoning or drug overdose can be presumed to be reasonably complete because a patient with a life-threatening poisoning or drug overdose is likely to be admitted to a hospital and included in this dataset regardless of health insurance status. Another strength of HDDS data is that, like death certificate vital statistics, the data are comprehensive for the states in which they are collected. Nonetheless, HDDS is not national and universal. As of 1998, 42 states collected such data. However, the consistency of E-coding was more variable, with only 36 states collecting some ICD-9 E-code data as part of their HDDS; less than half the states mandated such coding (American Public Health Association, 1998). To the extent that poisoning is a comorbid condition displaced within a longer list of diagnoses, it may not be captured in summary data. Iatrogenic causes of medication toxicity may be preferentially down-graded or obscured in such records.
The HDDS data exclude cases treated in an emergency department and discharged. Hospital-based emergency department data systems (HEDDS) exist, but they are limited. As of 1998, only 12 states had these systems in place. However, additional states reported plans to add similar systems. Once again, E-coding in these systems is variable. The data collection systems do not include freestanding urgent care centers or other outpatient treatment centers. As of 1998, only four states had a non-emergency department, statewide outpatient data system. The extent to which HDDS and HEDDS data are publicly available with minimal or nominal charge and the lag time between data collection and public data access vary by state. The potential limitations of ICD-9 nosology apply to HDDS and HEDDS data and are likely to be magnified by E-code deficiencies. Both the HDDS and HEDDS datasets are comparable to vital statistics death certificate data in that they are meant to capture all of the eligible events within the geographic areas they cover.