Symptom-Score des Sterbeprozesses zur Prognosebeurteilung des Sterbeprozesses bei Bewohnern in Pflegeheimen

Around 68% of all people in Austria die in hospital or in nursing homes. The nursing demands and human expectations placed on care staff during this time cannot be neglected. In order to meet the complex demands of this last phase of life, it seems necessary to identify the beginning of the dying process more correctly.

The aim of the study was to more accurately identify the beginning or the stage of the dying process of residents in long-term care and to enable the nursing staff to classify residents into groups according to the expected lifespan with the help of a symptom score card. In each group at least 60% of the residents should be classified correctly.

In order to establish a prognosis of the duration of the remaining lifetime, the residents’ care documentation was verified and accordingly several parameters were collected. In a statistical analysis the collected data were then retrospectively and anonymously evaluated. As a statistical methodology a random forest model was used and a supporting tool using artificial intelligence was developed. The resulting model selected 16 out of the 31 available variables from the care documentation and 3 groups were established according to the remaining lifetime (≤ 1 week, ≤ 8 weeks, > 8 weeks).

By using this model a correct prognosis for the last 8 weeks before death of all residents can be made in 74% of the cases.

With the help of the model which was developed based on the random-forest model, the start of the last phase of life of a large number of nursing home residents can be identified more correctly. Thus, a larger number of residents can expect to enjoy a better quality of life in this phase of life.