Surgery has curative potential in a proportion of patients with esophageal cancer, but is associated with considerable perioperative risks.
Dr Ewout Zetterberg and colleagues from the Netherlands developed and validate a simple risk score for surgical mortality that could be applied to administrative data.
The team analyzed 3592 esophagectomy patients from 4 cohorts.
The researchers applied logistic regression analysis to predict mortality occurring within 30 days after esophagectomy.
This analysis was done for 1327 esophageal cancer patients older than 65 years of age diagnosed between 1991 and 1996.
The team obtained the data from the linked Surveillance, Epidemiology and End Results-Medicare database.
|Surgical mortality in the 4 cohorts ranged from 4% to 11%|
|Journal of Clinical Oncology|
A simple score chart for preoperative risk assessment of surgical mortality was developed and validated on 3 other cohorts.
The team assessed 714 Surveillance, Epidemiology and End Results-Medicare patients diagnosed between 1997 and 1999.
A further 349 patients from a population-based registry in the Netherlands diagnosed between 1993 and 2001 was evaluated.
The researchers also analyzed 1202 patients from a referral hospital in the Netherlands diagnosed between 1980 and 2002.
The researchers found that surgical mortality in the 4 cohorts was 11%, 10%, 7%, and 4%, respectively.
Predictive patient characteristics included age, and comorbidity such as cardiac, pulmonary, renal, hepatic, and diabetes.
The team noted that preoperative radiotherapy or combined chemo radiotherapy were predictors of mortality.
And a relatively low hospital volume also predicted mortality.
At validation, the simple score showed good agreement of predicted risks with observed mortality rates, but low discrimination.
Dr Steyerberg's team concluded, “A simple risk score combining clinical characteristics along with hospital volume to predict surgical mortality after esophagectomy from administrative data may form a basis for risk adjustment in quality of care assessment.”