Ƶ

Humans Beat Stroke Scales for ICH Prognosis

Ƶ MedicalToday

This article is a collaboration between Ƶ and:

PHILADELPHIA -- Physicians and nurses in stroke units more accurately predicted outcomes in patients with intracerebral hemorrhage (ICH) compared with common prognostic scales, a researcher reported here.

Attending physicians' and nurses' predictions of future modified Rankin Scale (mRS) for 100 ICH patients they were treating, made within 24 hours of event onset, correlated with actual 3-month mRS scores with Spearman r values of 0.81 and 0.72, respectively, according to David Y. Hwang, MD, of Yale University.

Action Points

  • Note that this study was published as an abstract and presented at a conference. These data and conclusions should be considered to be preliminary until published in a peer-reviewed journal.
  • Note that this prospective cohort study demonstrated that physician and nurse predictions of outcome in patients with intracerebral hemorrhage were superior to the predictions of two commonly-used stroke scales.
  • Be aware that the stroke scales are designed to be quickly and easily implemented across multiple centers and would not be expected to outperform individualized prognostication.

These values were significantly (P<0.02) higher than the r values of 0.55 and 0.46 (absolute value) for ICH and scores, respectively, Hwang told attendees at the American Academy of Neurology's annual meeting.

The ICH Score system was first developed in 2001 and updated in 2009. It rates patients on a scale of 0 to 6 mainly on the basis of event characteristics: the patient's Glasgow Coma Score, the ICH volume, and whether the bleed site was intraventricular or infratentorial, along with whether the patient was older than 80. Higher scores reflect a greater likelihood of poor outcome.

FUNC (not an abbreviation) is similar but was designed more specifically for predicting recovery of functional independence. Its components include Glasgow score, three categories of ICH volume and age, bleed location as lobar, deep, or "other," and the presence of pre-event cognitive impairment. Higher FUNC score indicates a lower likelihood of poor outcome.

"Early clinical judgments of attending physicians and nurses are more accurate than the ICH or FUNC score for outcome prediction," Hwang said, presumably because the human professionals took factors into account that are not included in either scale.

The researchers did determine that the doctors' and nurses' knowledge of the care patients were likely to receive -- especially for those with extremely poor prognoses, who would most likely receive palliative care only -- was not responsible for their better performance compared with the scales.

When patients for whom only palliative care was recommended were excluded from the analysis, the human professionals still significantly outperformed the scales, Hwang reported.

Session moderator , of Columbia University in New York City, told Ƶ that these scales do perform a valuable clinical function, but it's not surprising that trained professionals would do better.

He said the scales were designed to be fast and easy-to-calculate aids to prognosis, and can't account for all the factors that affect outcomes.

For example, the scales don't include measures of patients' overall pre-event physical condition. But Marshall said it was likely that many of the professionals used the scales as part of their judgment.

Hwang said the participating professionals were asked about the factors they took into account in reaching their prognostic judgments, but analysis of those data had not been completed yet.

Study Details

For the current analysis, Hwang and colleagues piggybacked on an ongoing study called Ethnic/Racial Variations of Intracerebral Hemorrhage (ERICH), a large observational study being conducted in five major academic medical centers.

When a patient was accepted into ERICH, Hwang and colleagues sought to track down one physician and one nurse treating the patient within 24 hours of admission to predict the patient's mRS score 3 months later.

Hwang said ERICH was ideal for the study because it seeks to enroll ICH patients very quickly after presentation. Staff at the participating centers scan new admissions frequently throughout the day to find incoming ICH patients eligible for inclusion.

Even so, out of 405 ERICH patients potentially available for the study, the investigators could not obtain physician and nurse predictions of outcome within 24 hours in more than half. Other reasons kept 87 others out of the analysis, leaving 100 for which predictions were obtained and for which 3-month mRS scores were known.

Patients were fairly typical of those experiencing ICH: mean age was 67, the location was deep in about half and lobar in one-third, and the hemorrhage volume in 70% of cases was less than 30 mL and exceeded 60 mL in 13%.

Of the 100 physicians, 75 were attendings and 25 were trainees. About 70% of the physicians and the nurses had a neurology specialization, Hwang said.

He noted that a potential confounder was that the treating professionals would probably know when palliative care would be recommended, which would essentially dictate the outcome.

But excluding the 18 patients for which that was the case did not change the pattern, although it did reduce the correlation coefficients slightly.

For physicians and nurses, the r values after excluding such patients for the predictions versus actual 3-month mRS scores were 0.78 and 0.66, respectively. ICH and FUNC correlations were 0.48 and 0.37 (absolute value), respectively.

Disclosures

The analysis was funded by the National Institutes of Health. Study authors declared they had no relevant financial interests.

Primary Source

American Academy of Neurology

Hwang D, et al "Subjective judgments of physicians and nurses are more accurate than formal clinical scales in predicting functional outcome after intracerebral hemorrhage" AAN 2014; Abstract S45.006.