Premature discharge of patients from ICU increases mortality

The pressure on intensive care beds has led to the discharge of patients before clinically indicated. It has been shown that discharging patients with high TISS scores increases mortality [1], and it has been stated that 'premature' discharge is likely to worsen outcome [2]. We postulated that premature discharge would increase mortality, and that this would be independent of the different TISS scores that were likely to occur.

All ICU patients discharged alive to the ward during 1997-1999, whose reason for discharge had been recorded by the ICU consultant as fully fit or premature were studied. We excluded those discharged for palliative care. We compared the groups APACHE II, admission risk of death, length of ICU stay and last TISS using ANOVA. We then assessed the relative risk of hospital mortality and investigated the association between TISS on day of ICU discharge, reason for discharge and mortality using ordinal regression analysis. 552 patients were identified (Premature Group 145, Fully Fit Group 407). ANOVA identified a significant difference in last TISS, but no significant difference in APACHE II, risk of death or ICU length of stay (Table 1). The hospital mortality was greater in the premature group (relative risk 2.1, 95% CI 1.3-3.5). Both last TISS and discharge reason were found to be independent indicators of hospital mortality (Table 2).

Table 1 ICU data for patients

Table 2 Ordinal regression for hospital mortality using last TISS and premature discharge

Patients who leave ICU before they are considered fit for discharge are twice as likely to die, despite minimal difference in their risk of death on ICU admission. Although they are receiving more care when they are discharged the excess mortality in those discharged prematurely is equivalent to that associated with a further increase in last TISS of 14 points.