Publication details
Epidemiology Stroke Program in the Czech Republic – Is the National Registry of Hospitalized Patients Valid for Further Epidemiological Research?
| Authors | |
|---|---|
| Year of publication | 2014 |
| Type | Conference abstract |
| Citation | |
| Description | Objectives: Stroke is considered as a major cause of mortality and morbidity in East-Central Europe but valid epidemiological data are not available. As the first step to evaluate stroke epidemiology in the Czech Republic (CR) we validated the National Registry of Hospitalized Patients (NRHOSP). Methods: NRHOSP is a nationwide registry, prospectively collecting information on each hospitalization in the CR since 1960. We randomly selected 10 hospitals and then 50 patients from each hospital admitted in 2011 stratified according to stroke diagnosis (ICD 10 codes I60, I61, I63, I64 and G45). Discharge summaries from hospitalization were reviewed by two independent reviewers and compared with NRHOSP for accuracy of discharge diagnosis. Any disagreements between reviewers and NRHOSP were adjudicated by a third independent reviewer. Results: Of 500 requested discharge summaries, 484 were available for validation (response rate 97%). Stroke or TIA diagnosis (ICD code I6) recorded in NRHOSP was confirmed in 303 cases (62.6%, 95% CI: 59.6–65.7%). Of 385 cases coded in NRHOSP with stroke diagnosis only (I60, I61, I63 and I64) there were 326 cases (84.7%, 95% CI: 80.7–87.9%) confirmed with any of the stroke diagnoses (I60, I61, I63 and I64). Of these 326 cases, 255 (66.2%, 95% CI: 61.3–70.8%) were confirmed with the original stroke type diagnosis recorded in NRHOSP. With analysis by stroke type ischemic stroke I63 was confirmed in 80 of 98 cases (81.6%, 95% CI: 74.0–89.3%), subarachnoidal hemorrhage I60 was confirmed in 88 of 97 cases (90.7%, 95% CI: 85.0–96.5%), and intracerebral hemorrhage I61 was confirmed in 86 of 95 cases (90.5%, 95% CI: 84.6–96.4%). TIA G45 was identified in 48 of 99 cases (48.5%, 95% CI: 38.6–58.3%). The single most important reason for disagreement between discharge summaries and NRHOSP was incorrect coding of unspecified stroke I64, which was reported to NRHOSP (in 95 cases) but in fact patients had specified stroke type (in 63 cases, 66.3%), TIA (in 10 cases, 10.5%) or another non-cerebrovascular diagnosis (in 21 cases, 22.1%). Conclusions: The accuracy of coding of ischemic stroke I63, subarachnoid hemorrhage I60 and intracerebral hemorrhage I61 in NRHOSP were high. The data suggest that with appropriate consideration of stroke cases included in the unspecified stroke I64 and TIA G45 categories, the NRHOSP database can be utilized to provide high level epidemiological data regarding the occurrence of stroke in the Czech Republic. The data also suggest that the epidemiology of TIA (G45) in a population cannot be accurately assessed using data from an administrative database. |