Can the wrong statistic be bad for health? Improving the reporting of door-to-needle time performance in acute myocardial infarction

Abstract

Current acute myocardial infarction (AMI) guidelines call for reperfusion to be given to all eligible patients within a set time interval after hospital arrival, yet current hospital performance benchmarks are based on the median door-to-intervention time among treated patients. Our objective is to compare hospital performance rankings when door-to-needle time (DNT) is measured at the current benchmark (median < or = 30 minutes) versus those obtained with more stringent benchmarks common for other AMI treatments. A secondary analysis of data from the EFFECT study from 52 small, community and teaching hospitals in Ontario. All Ontario hospital corporations that treated > or = 30 patients with AMI from 1999 to 2001 participated. The charts of approximately 125 patients with AMI per hospital were reviewed; median and 85th percentile DNTs were then calculated for patients with ST-elevation AMI given thrombolysis at each site along with the proportion of patients thrombolysed within the recommended time. Hospitals were then ranked according to each indicator. Data were obtained on 1,578 patients given thrombolytic drugs at 52 hospitals. The median and 85th percentile DNTs were 37 and 82 minutes, respectively; the proportion of patients treated in < or = 30 minutes ranged from 8.5% to 60%. Hospitals that achieved a median DNT of < or = 30 minutes treated 40% to 50% of their patients outside that time frame. The ranks of the top 15 median DNT hospitals changed substantially when re-ranked according to the 85th percentile (average change in rank -16, range +6 to -40). If DNT improved such that a 30-minute median target was achieved, the estimated reduction for the average patient would be 13 minutes versus a 43-minute reduction if the 85th percentile target was achieved. Hospitals that achieve a 30-minute median DNT benchmark still treat 40% to 50% of their patients outside the recommended time, which is not consistent with current AMI treatment guidelines. Door-to-needle time for the average patient would be up to 43 minutes faster if the DNT target was achieved at the 85th percentile.