Physicians and Nurses May Predict ICH Outcomes Better Than Clinical Scales Do

SAN DIEGO—Health care providers’ clinical judgments may be more accurate predictors of outcome for patients with intracerebral hemorrhage (ICH) than formal clinical scales, according to research presented at the 2014 International Stroke Conference. Data indicate that predictions made within 24 hours of patient admission by physicians and nurses correlate more closely with patient outcome at three months than either the ICH score or FUNC score.

When the researchers were interpreting the data, they considered the possibility that the physicians and nurses had an advantage over the clinical scales because they were members of the treatment team and thus knew in advance which patients might be treated with comfort measures alone. But when the investigators removed from the analysis 18 patients for whom providers indicated that they likely would recommend comfort care, the subjective predictions still correlated with three-month outcome better than the results from either clinical scale did, said David Y. Hwang, MD, Assistant Professor of Neurology at the Yale School of Medicine in New Haven, Connecticut. All relationships were statistically significant and remained true for an analysis that included only the survivors in the cohort.

David Y. Hwang, MD

Clinicians Predicted Modified Rankin Scores at Three Months
Accurate early prognoses for patients with ICH guide therapeutic decisions and influence the counseling that clinicians give to patients and their families. To compare the accuracy of the ICH score, the FUNC score, and the subjective clinical judgment of physicians and nurses at predicting functional outcome at three months, Dr. Hwang and colleagues conducted a prospective, observational study involving two cohorts. The first cohort comprised adult patients with ICH who already were being enrolled in a larger study called the Ethnic/Racial Variations of Intracerebral Hemorrhage (ERICH). The second cohort comprised attending physicians, physician trainees, and nurses taking care of these patients. For each enrolled patient with ICH, one physician and one nurse caring for him or her were also enrolled in the study.

The investigators asked the clinicians to predict the three-month modified Rankin scale score for each enrolled patient within 24 hours of patient admission. The physicians and nurses also indicated whether they would recommend comfort measures to the patient’s surrogate decision maker, based on the estimated prognosis. Using information and imaging data collected in the larger study, Dr. Hwang and colleagues calculated admission ICH and FUNC scores for each patient. The researchers also obtained actual three-month modified Rankin scale scores for each patient. They calculated Spearman’s rank correlations for the clinical predictions, the ICH score, and the FUNC score with regard to actual, blinded three-month outcomes.

More than 400 patients were eligible for the study, but approximately three-fourths of them were excluded from the final analysis, mostly because of the practical challenge of obtaining an early outcome prediction from both a nurse and a physician. The analysis was based on data for 100 patients. The average patient age was 66, and 64% of patients had a Glasgow Coma Scale score between 13 and 15. About half of patients had bleeds that were deep, and 71% of bleeds had volumes smaller than 30 cm³.

Mortality Was Similar to That Seen in Prior Large Series of Patients With ICH
Spearman’s rank correlations for predictions made by the attending physicians and nurses were significantly higher than the correlations for the ICH or the FUNC scores. “Of note, all P values for individual comparisons between attendees and nurses versus either the ICH score or the FUNC score are less than 0.02,” said Dr. Hwang.

One limitation of the study was that participating physicians and nurses were not formally trained in the modified Rankin scale before the study began, said Dr. Hwang. The prediction form that physicians and nurses used, however, contained details about the modified Rankin scale to aid participants. In addition, the majority of patients appeared relatively well and had good exams and small bleeds. Yet “the overall mortality at three months in our cohort was 35%, which is comparable to [that of] other large published series of ICH patients,” observed Dr. Hwang.

“The purpose of having attending physicians and nurses that were not blinded, that were members of the treatment team, was to try to mimic real life,” he added. “In real life, we … are tasked with the responsibility of trying to provide accurate prognostication to patients and their families. The idea was to try to test the accuracy of these real-life, bedside decisions versus the ICH score and FUNC score, two scales that we have grown to rely on.”