Abstracts

Health services research in epilepsy is in its infancy in many parts of the world. This type of research is crucial in understanding provisional health services in epilepsy, treatment gaps, and determinants of variations encountered. Whether the right patients are receiving the right interventions at the right time requires research of this nature. Administrative databases are a rich resource that has been used in many areas but rarely in epilepsy. ICD9, and more recently ICD10 codes are utilized worldwide in administrative databases. There is little information on the validity of ICD9 codes for epilepsy and no information on ICD10 codes. We undertook a before and after comparison of ICD9 and ICD10 epilepsy coding in inpatients in a large health region. All discharge diagnoses in Canada are ICD coded and captured in a central repository. ICD9 codes were used in our region until 2000. Thereafter ICD10 codes were adopted. We obtained ICD codes epilepsy- related, in any sequence (whether most responsible diagnosis for admission or not) for 2000 (ICD9) and 2004 (ICD10), in five inpatient facilities within the entire Calgary Health Region (population 1.4 million). We grouped ICD codes into generalized, localization-related, status epilepticus and other. We examined the distribution of each group within the two coding systems, both as a most responsible diagnosis and as other diagnoses. The following ICD 9 codes were used: generalized epilepsy (345.0, 345.1, 780.4, 333.2), localization-related epilepsy (345.4, 345.5), status epilepticus (345.3, 345.7), other epilepsy (345.8, 345.9). For ICD 10 the following codes were used: generalized epilepsy (G40.3, G40.4, G40.6, G40.7), localization-related (G40.0, G40.1, G40.2), status epilepticus (G41.0, G41.2, G41.8, G41.9, G40.5), other epilepsy (G40.8, G40.9). In the absence of overt changes in availability of inpatient services, ICD9 captured 968 patients and ICD10 captured 625 per year.
Table 1 In the absence of expected changes in the patient population or rate of admissions, substantial discrepancies occurred between types of epilepsy and numbers of epilepsy captured by the two coding systems. ICD10 coding seems better at capturing localization-related epilepsy and status epilepticus as well as having fewer patients in the non-specific (other epilepsy) category. This difference was apparent both for most responsible diagnosis and for other diagnoses. Before ICD codes can be used extensively in health services research, further research should focus on their accuracy based on comparison with clinical diagnoses, in addition to exploring variables that explain differences between systems.[table1]