Table 3.
Number in the invasive group | Number in the non-invasive group | Hazard ratio (95% CI) | p value | |
---|---|---|---|---|
Primary analysis | ||||
Multivariable* plus PS adjustment, with deaths within 3 days of peak troponin concentration assigned to a treatment group based on PS† | 686 | 892 | 0·68 (0·55–0·84) | <0·0001 |
Sensitivity analyses | ||||
Crude, excluding deaths within 3 days of peak troponin concentration | 655 | 845 | 0·38 (0·32–0·46) | <0·0001 |
Multivariable* plus PS adjustment, excluding deaths within 3 days | 655 | 845 | 0·56 (0·45–0·70) | <0·0001 |
IPTW, excluding deaths within 3 days | 655 | 845 | 0·66 (0·49–0·88) | 0·0050 |
Multivariable* plus PS adjustment, restricted to patients with PS in the 25th and 75th percentiles, excluding deaths within 3 days | 411 | 528 | 0·56 (0·44–0·71) | <0·0001 |
Multivariable* plus PS adjustment, with deaths within 3 days of peak troponin concentration assigned to the non-invasive group | 655 | 923 | 0·50 (0·40–0·61) | <0·0001 |
Multivariable* plus PS adjustment, with deaths within 3 days of peak troponin concentration assigned to the invasive group | 733 | 845 | 1·05 (0·86–1·27) | 0·65 |
IPTW=inverse probability of treatment weighted. PS=propensity score.
Adjusted for age, interhospital transfer, creatinine and haemoglobin concentration, family history of ischaemic heart disease, hypercholesterolaemia, hypertension, abdominal aortic aneurysm, angina, aortic stenosis, cardiogenic shock, heart failure, previous myocardial infarction, supraventricular tachycardia, ventricular fibrillation, ventricular tachycardia, acute renal failure, urinary tract infection, interstitial lung disease, obstructive lung disease, other lung disease, ischaemic stroke, Parkinson's disease, anxiety, gastric ulcer, metabolic disorder, bowel incontinence, dementia, history of falls, mild cognitive impairment (no dementia), need for mobility assistance, and speaking difficulty.
57 of the 77 deaths in patients not invasively managed within 3 days of the peak troponin concentration had eligible propensity scores and were assigned to treatment groups using multiple imputation. Across the imputed datasets, on average, ten were included in the invasive management group and 47 were included in the non-invasive management group.