Ten studies were included in the review (n=17,265) and eight studies were included in the meta-analysis. All studies recruited a representative patient spectrum in an unbiased fashion and assessed the predictor variable blinded to outcome. Seven studies reported accurate definitions of outcomes. Three studies gave an explicit interpretation of the risk score by clinicians in practice blinded to outcome. Eight studies reported adequate follow-up (<10% loss). Only one study reported assessing the outcome blinded to predictor variables.
Sensitivity decreased as the TIMI threshold increased, from 97.2% (95% CI 96.4% to 97.8%) when the cut off was more than zero to 33.2% (95% CI 31.3% to 35.3%) when the cut off was more than 4. Specificity increased as the TIMI threshold increased, from 25.0% (95% CI 24.3% to 25.7%) when the cut off was more than zero to 96.6% (95% CI 96.3% to 96.9%) when the cut off was more than 4.
Positive likelihood ratio increased as the TIMI threshold increased, from 1.30 (95% CI 1.28 to 1.31) when the cut off was more than zero to 9.70 (95% CI 8.7 to 10.8) when the cut off was 4. Negative likelihood ratio increased as the TIMI threshold increased, from 0.11 (95% CI 0.09 to 0.15) when the cut off was more than zero to 0.69 (95% CI 0.67 to 0.71) when the cut off was more than 4. Where reported, highly statistically significant heterogeneity was observed.
There was significant interaction between prevalence and diagnostic accuracy at TIMI thresholds of more than zero and more than 1. Meta-regression showed a strong linear relationship between TIMI risk score and cumulative incidence of cardiac events (p<0.001).