Development of a decision analytic model (DAM) to support decision-making and risk communication for thrombolytic treatment in acute stroke care

Background: Thrombolysis with iv rt-PA has improved the prognosis for acute ischaemic stroke. Decision support during the hyperacute stroke period is needed to expedite appropriate clinical assessment of eligibility for rt-PA, and communication of risks and benefits to patients/families. Methods: A DAM was constructed to establish the likely balance of benefits and risks of thrombolytic treatment in individual patients. Probability of independence (mRS 0-2) three months post-stroke was based on a predictive equation reported in the literature (Stroke-Thrombolytic Predictive Instrument [STPI], derived from large scale trials) calibrated using data from stroke patients in the SITS-UK database. Probabilities for death at three months were derived from regression analyses of SITS-UK data; whereas, probability of symptomatic intracerebral haemorrhage (SICH) used an equation derived from SITS data. Results: The DAM expresses probabilities for short-term (SICH, death, independence and dependence at 6-months) and long-term outcomes (QALYs), with and without rt-PA, as a function of 11 patient characteristics (age, gender, diabetes, previous stroke, NIHSS score, systolic BP, onset time to treatment, weight, aspirin use, blood glucose and signs of current infarction on imaging). Adding blood glucose and signs of current infarction to the S-TPI predictive equation increased prediction accuracy (area under the curve increased from 0.728 to 0.741). The DAM identifies sub-groups of patients with a different balance of risks and benefits from thrombolytic treatment, including no benefit and likely harm. Conclusion: Outcomes generated by the DAM show improved discrimination between stroke patients who may and may not maintain their independence. The DAM has been embedded within a computerised tool for supporting clinical decision making on rt-PA and risk communication (and where appropriate engagement of patients/family members in decision-making) within the hyperacute stroke period.