The Hemorrhage After Thrombolysis (HAT) score is a 0–5 point tool developed to estimate the risk of intracerebral haemorrhage after standard-dose IV alteplase for acute ischaemic stroke.

Original evidence

Derived using 302 IV t-PA-treated patients from the NINDS trials, with complete data for diabetes or admission glucose, pretreatment NIHSS, and visible hypodensity on baseline non-contrast CT.
Internally validated in a separate cohort of 98 consecutive IV t-PA patients from Beth Israel Deaconess Medical Center (2001–2007), all treated within 3 hours of onset with standard 0.9 mg/kg dosing, baseline CT, and CT or MRI at 24–48 hours for haemorrhage assessment.
Symptomatic intracerebral haemorrhage (SICH) was defined as any ICH on follow-up imaging with an NIHSS increase of ≥4 points from baseline; haemorrhage with final fatal outcome (HFFO) was death during the admission after ICH.

Risk by HAT score in combined cohorts (n = 400)

HAT score	Symptomatic ICH (SICH)	Haemorrhage with final fatal outcome (HFFO)
0	≈2%	0%
1	≈5%	≈3%
2	≈10%	≈7%
3	≈15%	≈6%
≥4	≈44%	≈33%

Across the same cohorts, higher HAT scores were also associated with worse 3‑month functional outcome (higher modified Rankin Scale scores) in both alteplase and placebo groups, with slightly better prognostic discrimination in the t-PA-treated patients.

Evidence appraisal and limitations

In the combined NINDS and Beth Israel cohorts, the c-statistic for predicting SICH was ~0.74 and for HFFO ~0.79, indicating moderate discrimination rather than precise individual prediction.
A small retrospective community-hospital series of 39 alteplase-treated patients found that the HAT score predicted ICH with an AUC of 0.66, in-hospital mortality with an AUC of 0.67, and favourable functional outcome (mRS <2) with an AUC of 0.77, suggesting only modest performance outside trial settings.
In a larger trial population, both the HAT and Multicentre Stroke Survey risk scores were associated with SICH but showed limited discriminatory ability, similar to other SICH risk tools.
Later work has proposed multiple alternative scores (e.g. SEDAN, DRAGON, GRASPS, SITS-SICH) and newer prediction models that incorporate similar predictors; overall, most scores offer only moderate separation of low- and high-risk patients and are best used as adjuncts rather than sole decision-makers.
The score was developed for adults receiving standard-dose IV alteplase within 3 hours of onset using early non-contrast CT, and absolute haemorrhage risks from these cohorts are likely to overestimate risk in contemporary practice where broader selection criteria, advanced imaging, and systems of care have reduced overall SICH rates.

Use the HAT score to inform discussion and monitoring after IV alteplase, but do not use it alone to deny otherwise guideline-concordant thrombolysis or thrombectomy.

The primary source or current guidance provides the model definition, source population and supporting evidence.

HAT Score for Haemorrhage After Thrombolysis