Differences in process management and in-hospital delays in treatment with iv thrombolysis

Abstract

Objectives: Rapid initiation of intravenous thrombolysis improves patient's outcome in acute stroke. We analyzed inter-center variability and factors that influence the door-to-needle time with a special focus on process measurements in all Austrian stroke units.

Methods: Case level data of patients receiving intravenous thrombolysis in the Austrian Stroke Unit Registry were enriched with information of a structured questionnaire on center specific process measures of all Austrian stroke units. Influence of case and center specific variables was determined by LASSO procedure.

Results: Center specific median door-to-needle time ranged between 30 and 78 minutes. Between April 2004 and November 2012, 6246 of 57991 patients treated in Austrian stroke units with acute ischemic stroke received intravenous thrombolysis. An onset-to-door time >120 minutes, patients with total anterior circulation stroke, recent year of admission, patient transportation with ambulance crew and emergency physician, the use of point of care tests reduced the door-to-needle time, whereas onset-to-door ≤ 60 minutes, unknown onset-to-door, patients with an NIHSS ≤ 4 or posterior circulation stroke, initial admission to a general emergency department, a distant radiology department, primary imaging modality other than plain CT and waiting for the lab results were associated with an increase in door-to-needle time. Case level and center specific factors could explain the inter center variability of door-to-needle times in 31 of 34 stroke units in Austria.

Conclusions: In light of our results it seems crucial that every single stroke center documents and critically reviews possibilities of optimizing practice strategies in acute stroke care.

Figure 1. Inter center variability of the door-to-needle times in the 34 stroke units in Austria (box-plots).

Figure 2. Multiple regression model of case level and center specific factors on the door-to-needle time in Austria.

The model contains 18 coefficients (including the intercept) and is based on 5858 observations (adjusted R2 0.14). Since the target variable is log(DNT) the coefficients are not additive in terms of the DNT, but the exp(coefficients) are multiplicative factors in relation to the reference value exp(Intercept).