Predicting early death in older adults with cancer

Predicting early death after a comprehensive geriatric assessment (CGA) is very difficult in clinical practice. The aim of this study was to develop a scoring system to estimate risk of death at 100 days in elderly cancer patients to assist the therapeutic decision.

This was a multicentric, prospective cohort study approved by an ethics committee. Elderly cancer patients aged older than 70 years were enrolled before the final therapeutic decision. A standardised CGA was made before the treatment decision at baseline. Within 100 days, event (death), oncologic and geriatric data were collected. Multivariate logistic regression was used to select the risk factors for the overall population. Score points were assigned to each risk factor using the β coefficient. Internal validation was performed by a bootstrap method. Calibration was assessed with the Hosmer-Lemeshow goodness of fit test and accuracy with the mean c-statistic.

One thousand fifty patients (mean age: 82 years) joined the study from April 2012 to December 2014. The independent predictors were metastatic cancers (odds ratio [OR] 2.5; 95% confidence interval [CI], [1.7-3.5] p<0 .001); gait speed<0.8 m/s (OR 2.1; 95% CI [1.3-3.3] p=0.001); Mini Nutritional Assessment (MNA) < 17 (OR 8; 95% CI; [3.7-17.3] p<0.001), MNA ≤23.5 and ≥ 17 (OR 4.4; 95% CI, [2.1-9.1) p<0.001); performance status (PS) > 2 (OR 1.7; 95% CI, [1.1-2.6)] p=0.015) and cancers other than breast cancer (OR 4; 95% CI, [2.1-7.9] p<0.001). We attributed 4 points for MNA<17, 3 points for MNA between ≤23.5 and ≥ 17, 2 points for metastatic cancers, 1 point for gait speed <0.8 m/s, 1 point for PS > 2 and 3 points for cancers other than breast cancer. The risk of death at 100 days was 4% for 0 to 6 points, 24% for 7 to 8 points, 39% for 9 to 10 points and 67% for 11 points.

To our knowledge, this is the first score which estimates early death in elderly cancer patients. The system could assist in the treatment decision for elderly cancer patients.