• Patient safety
• Quality improvement
• Quality improvement methodologies

Recently, Provenzano et al1 found that an electronic tool collecting real-time clinical information directly from frontline providers was both feasible and useful to evaluate inpatient deaths.

These findings concur with our evaluation of the preventability of death using a simple electronic evaluation tool in our 46-bed adult intensive care unit (ICU).

From September 2010 to September 2011, an email was sent to the attending intensivist each time a patient died in our intensive care including two questions: “Was this death preventable? If yes, what was the cause of preventability?” The definition of preventable mortality was provided using three criteria: the illness was survivable, care was suboptimal and suboptimal care was related to death. No reminding emails were sent. In addition, the patient charts of all cases were retrospectively reviewed by two ICU nurses and a physician.

A total of 306 patients (9.9%) died. Acute Physiology and Chronic Health Evaluation (APACHE) IV standardised mortality rate was 0.77. In 48 of these deceased patients, the APACHE IV-based mortality risk was below 20%. Response rate was 92% and 47 deaths (15%) were reported to be potentially preventable. Large interindividual variations between the intensivists (n=24) were observed. Response varied between 65% and 100%, and preventable death judgements varied from none to 66%. When using blinded chart review death was judged potentially preventable by the nurses and physician in 7%, 11% and 18%, respectively.

Similar to Provenzano et al, we also found poor agreement between the preventability ratings from frontline intensivist reviews compared with blinded chart review.2 In 21 cases (45%) in which the intensivist scored a preventable death, all three reviewers scored these non-preventable. This might be partly explained by additional information on each patient's individual circumstances that cannot be easily deduced from patients’ charts. Using APACHE IV as selection criterion for in-depth evaluation is insufficient while analysis of patients with an APACHE IV-based risk of mortality below 20% showed that only four of these deaths (8.3%) were considered potentially preventable.3

Preventability of death evaluation of all inpatient deaths is required either for quality improvement and/or by regulatory authorities.

A quick and efficient method with high response rates from frontline providers is feasible and may provide useful information for quality improvement.4 However, large interindividual variations in response and judgement exist, and, therefore, this method apparently is insufficient for benchmarking.