A preliminary study of the reliability of immediate vs delayed interviews of cardiac arrest witnesses

INTRODUCTION: Methods to characterize the interval between a collapse from cardiac arrest until a 911 call is made have not yet been developed. OBJECTIVE: To determine the concordance of cardiac arrest data obtained by two methods: an immediate nurse interview of out-of-hospital cardiac arrest (OHCA) witnesses, and a follow-up phone interview performed two weeks later. METHODS: This was a prospective study of OHCA witnesses dating from January 1997 to May 1998. Witnesses were briefly interviewed at the time of emergency department presentation, and two weeks later a more lengthy structured phone interview was performed. The authors identified key data elements: 1) was the arrest witnessed? (Wit); 2) was CPR administered prior to EMS arrival? (BCPR); 3) was the first call placed to 911? (c911); and 4) was the estimated collapse to call interval <4 minutes? (ECCI). The analysis utilized Cohen's kappa statistic and Spearman's correlation coefficient. RESULTS: A convenience sample of 42 matched pairs of OHCA cases was analyzed. Kappa statistics for agreement between methods were: 1) Wit(kappa = 0.750), 2) BCPR(kappa = 0.892), 3) c911 (kappa = 0.892), and 4) ECCI(kappa = 0.571, Spearman's 0.528). CONCLUSION: There is good to excellent agreement between immediate and phone interview data retrieval methods. Phone interviews appear to yield data comparable to that with the more difficult and expensive, direct interview method.