Methodological challenges in the coding and adjudication of sudden deaths in a large simple trial with observational follow-up: the ziprasidone observational study of cardiac outcomes (ZODIAC)
ABSTRACT
Purpose
The Ziprasidone Observational study of car DIAC Outcomes (ZODIAC), a large simple trial comparing ziprasidone versus olanzapine in real-world use, showed no difference in risk of sudden death. Upon the request of the US Food and Drug Administration, 205 fatal events were readjudicated applying ICD-10 coding rules for sudden death.
Methods
A readjudication committee coded three domains (witness to death, time of symptom onset to death, and most likely cause of death) for use within algorithms consistent with ICD-10 rules. Relative risks (RR) and corresponding 95%CI were calculated for persons randomized to ziprasidone versus olanzapine, comparing 1-year incidence of sudden death, using multiple definitions.
Results
Data on symptom onset to death and diagnosis of specific cardiac arrhythmias required by the ICD-10 rules were often lacking. Sensitivity analyses were conducted to explore the impact of cases suggestive of cardiac origin but missing data required by ICD-10 sudden death codes. Overall, the readjudicated data matched the study's initial findings, with no significant difference in 1-year mortality between ziprasidone and olanzapine for sudden death not otherwise specified and sudden cardiac death (R96.0 or R96.1 or I46.1; RR = 1.11, 95%CI 0.45– 2.77).
Conclusions
After outcome readjudication, ZODIAC found no difference in the risk of sudden death among those randomized to ziprasidone versus olanzapine. However, unlike hospital-based studies, fatal events in general population studies often occur outside hospital and often lack the clinical detail needed for the exact determination of symptom onset and event. Epidemiological evaluations of sudden death need to consider the limitations of the available data. Copyright © 2011 John Wiley & Sons, Ltd.
