Predictors of stroke during 9-1-1 calls: opportunities for improving EMS response.

OBJECTIVE: Only 31%-52% of stroke calls are accurately identified by 9-1-1 dispatchers according to prior studies. Recognizing the time-dependent nature of acute stroke, better identification of stroke patients at the time of their 9-1-1 calls may allow an improved prehospital response. We sought to identify any words/phrases that 9-1-1 dispatchers could use to identify more stroke calls.
METHODS: Potential stroke calls were identified from emergency medical services run sheets, and the discharge diagnosis for each of these patients was obtained. The emergency medical services tapes were independently reviewed by two listeners who were blinded to the final diagnosis. Words/phrases previously associated with 9-1-1 stroke calls mentioned by the caller were recorded. Other pertinent words/phrases were also recorded. Using the final diagnosis of stroke as the gold standard, the sensitivity, specificity, and positive likelihood ratio of each word and phrase were calculated. Cohen's kappa was calculated to assess interrater agreement. Data were collected for runs from October 2003 to July 2004.
RESULTS: A total of 176 tapes were reviewed (40 strokes, 136 nonstrokes). The presence of at least one of four criteria predicted 80% of all stroke calls: the word "stroke," facial droop, weakness/fall, and impaired communication. All criterion elements had very good interrater agreement (kappa > 0.7). The word "stroke" was highly predictive of actual stroke (positive likelihood ratio, 2.27).
CONCLUSIONS: The majority of stroke patients in this study could be identified by 9-1-1 dispatchers if the caller reported any one of the following four complaints: stroke, facial droop, weakness/fall, or impaired communication.