ANN ARBOR, Mich. Each year, tens of thousands of people receive a dreaded diagnosis: intracerebral hemorrhage, or a "bleeding stroke." Caused by a burst blood vessel in the brain, ICH kills a quarter of patients in two days, and up to half of them within 30 days. And there's no approved specific medical treatment for it -- though people can recover with specialized hospital care.

Figuring out which ICH patients might survive if they receive aggressive treatment, and which will die or be severely disabled, challenges doctors every day.

Now, a new study from the University of Michigan Stroke Program suggests that the way those odds are calculated might be skewed.

It also lends credence to the idea that ICH patients might be victims of a "self-fulfilling prophecy": that their odds of survival may be made worse by the withholding of aggressive treatment based on an inaccurate calculation of their chances.

The study finds that ICH patients who had a do-not-resuscitate order issued in the first 24 hours after their stroke, or had care withdrawn or withheld in that time, were twice as likely to die as other ICH patients. It is published in the journal Neurology.

The difference in likelihood of death was independent of other factors typically used to predict ICH death risk, including coma score, age, gender, and the size of the bleeding area. The study involved 270 patients who were treated for ICH at seven community hospitals in Texas over a three-year period.

The fact that early limitations on patients' care were associated with such a large difference in mortality risk surprised the researchers. It's such a large effect that they say it should probably be considered when doctors use risk-calculation tools to predict the chance of death after ICH. Such tools are based on data from groups of past ICH patients, but none of the current tools take into account the level of care the patients received.