Ability of laypersons to estimate short time intervals in cardiac arrest

Study objective: Estimates of time intervals by bystanders are considered c ritical in cardiac arrest, and are often used in other disorders such as st roke and myocardial infarction. Because they have never been previously stu died, we sought to determine their accuracy. Methods: This study was performed by prospective collection of bystander es timates (made at the time of the arrest) of the time from calling 911 to th e arrival of urban fire department first responders, and comparison with ac tual measured response interval from computerized records, in all out-of-ho spital cardiac arrests from January 1996 through June 1998. Results: The fire department responded to 1,015 patients in cardiac arrest during the study period. First responders arrived before advanced life supp ort providers to 831 patients, who thus met study entry criteria. Bystander estimates were obtained in 497 of these 831 patients, who did not differ i n key characteristics from those lacking estimates. The bystander's average estimated fire department response interval was 5.6 minutes (95% confidenc e interval [CI] 5.2 to 5.9 minutes) and the actual measured interval to the patient's side from computer records was 6.1 minutes (95% CI 5.9 to 6.4 mi nutes). However, the median error of the bystander estimate (1.3 minutes) w as 32% of the median of the actual measured on-scene interval, and there wa s no correlation between the bystander estimates and the measured interval in individual cases (R less than or equal to 0.14), regardless of which int ervals were examined. Seventy-five percent of the bystander estimates erred by 20% or more. When bystanders estimated a response interval as excessive ly long, they were almost invariably wrong, but they also usually failed to identify intervals that actually were long. Conclusion: Although many diagnostic and research conclusions are based on interval estimates from laypersons, we found no correlation between estimat es and actual measured intervals in cardiac arrest. Current methodology may not be developed well enough to provide reliable data for research or qual ity assurance, and other clinical time estimates by patients and bystanders may be equally unreliable.