Estimating Long-Term Survival of Critically Ill Patients: The PREDICT Model
2008

Estimating Long-Term Survival of Critically Ill Patients: The PREDICT Model

Sample size: 11930 publication Evidence: moderate

Author Information

Author(s): Ho Kwok M., Knuiman Matthew, Finn Judith, Webb Steven A.

Primary Institution: University of Western Australia and Royal Perth Hospital

Hypothesis

Can a prognostic model accurately estimate long-term survival of critically ill patients?

Conclusion

The PREDICT model can estimate long-term survival probabilities for critically ill patients based on age, co-morbidities, and severity of illness.

Supporting Evidence

  • Age and co-morbidity are the most important determinants of long-term prognosis.
  • The model provides median survival time and long-term survival probabilities.
  • Discrimination performance of the model was assessed with a c-index of 0.757.
  • Patients with higher age and co-morbidities had worse survival outcomes.
  • The model is the first to estimate survival probabilities up to 15 years after critical illness.

Takeaway

Doctors can use a special tool to guess how long critically ill patients might live after their treatment, based on their age and health problems.

Methodology

This was a retrospective linked data cohort study involving critically ill patients who survived more than 5 days in an ICU.

Potential Biases

The model may not fully capture the impact of patients' wishes and quality of life on treatment decisions.

Limitations

The model's predictions should be considered average estimates and not used for individual patients; it may not account for all factors affecting survival.

Participant Demographics

The cohort included 11,930 critically ill patients, with a mean age of 53.8 years, and 62.8% were male.

Statistical Information

Confidence Interval

95% confidence interval 0.745–0.769

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1371/journal.pone.0003226

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication