Predicting the End of Life

Prognostication predictor scoring system for cancer under development

(RxWiki News) When cancer can no longer be effectively treated, palliative or comfort care begins. Knowing how long one has to live can help patients, their caregivers and families plan better. A new predictor model is being developed to do just that.

Researchers have developed a scoring method that can accurately and reliably predict how long an advanced cancer patient will likely live. The model determines whether the individual has days, weeks or months to live.

"Scoring system predicts survival time for advanced cancer patients."

Having a good idea of how long a seriously ill person will live helps both family and professional caregivers plan next steps. However, current methods for calculating survival time have been unreliable, often overly optimistic and subjective, according to study authors.

This study involved 1,018 patients who had incurable cancer, were no longer receiving medical treatment and had recently been referred to receive palliative care in the UK.

The team, led by Dr. Paddy Stone at St. George's, University of London, combined laboratory and clinical variables that are currently used to predict length of survival. All factors relating to the disease and the person's age, race and gender were accounted for.

Researchers then created two so-called prognostic scores that determined if patients were likely to live "days," defined as 0-13 days; “weeks” (seen as 14-55 days) or “months” (more than 55 days).

These scores were then compared with actual survival times and the predictions of clinicians.

Results showed that both scores were at least as accurate as an individual clinician's predictions. The scales were not, however, substantially more accurate than a multidisciplinary team prediction of survival.

Authors say further refinement is necessary before the system can become clinically available.

This research was published in the British Journal of Medicine. 

Review Date: 
August 29, 2011