Prediction of outcome in intensive care patients using endocrine parameters.

OBJECTIVE

To determine the ability of various endocrine parameters, measured at the time of intensive care unit (ICU) admission, to predict patient outcome.

DESIGN

Prospective, cohort study of patients requiring intensive care.

SETTING

The medical/surgical ICU at South Cleveland Hospital, UK and a medical/surgical ICU in a UK district hospital.

PATIENTS

A total of 260 consecutive patients requiring intensive care over a 2-yr period.

INTERVENTIONS

Patients were investigated within 1 hr of ICU admission by measuring plasma cortisol, serum thyroxine, triiodothyronine, and thyrotropin concentrations and by obtaining the Acute Physiology and Chronic Health Evaluation (APACHE II) score. Individual variables were compared between survivors and nonsurvivors.

MEASUREMENTS AND MAIN RESULTS

There were significant differences for each endocrine parameter between survivors and nonsurvivors (all p < .01). A multiple logistic regression analysis showed that only thyroxine, thyrotropin, and cortisol concentrations were independent predictors of outcome. An equation using these variables predicted outcome with 82% accuracy at the 0.5 cutoff point of the receiver operating curve. APACHE II scores predicted outcome with 72% accuracy at the same point on the receiver operating curve. Correct prediction of death was more frequent with the Endocrine Index than with APACHE II scores. Overall predictive power of the Endocrine Index, as measured by the area under the receiver operating curve, was 0.94 (95% confidence interval 0.91 to 0.96) vs. 0.85 (95% confidence interval 0.81 to 0.89) for APACHE II scores. Combining APACHE II scores and the endocrine parameters in a single index did not improve prediction (area under receiver operating curve = 0.94).

CONCLUSION

An endocrine prognostic index based on ICU admission measurements of thyroxine, thyrotropin, and cortisol concentrations is a superior discriminator of patient outcome than the APACHE II score.