Inclusion criteria in sepsis trials should include severity of organ dysfunction

Limitations of study entry criteria may be a crucial factor in the failure of immunomodulatory trials in sepsis. Several studies have used very broad inclusion criteria such as septic shock < 12 hours, sepsis syndrome, severe sepsis or more recently various combinations of SIRS criteria with at least one organ dysfunction. These criteria may include patients with a wide range of illness severity, reducing the sensitivity of studies using them.

Methods

Data analysis was performed for 5400 patients admitted to the Intensive Care Unit from 1994-1999. Patients fulfilling SIRS criteria were stratified according to the number of organ systems that had failed [1] and the duration of failure of those systems on the first day of SIRS. Mortality within the groups was compared using the chi-square test. Numbers of organs failed (and duration of those organ failures) at the onset of SIRS were compared in survivors and non-survivors of ICU using the Mann-Whitney U-test.

Results

3259 (60.4%) patients developed SIRS; 1077 (33%) died. Presence of temperature, WBC count, heart rate or respiratory criteria alone or in combination did not influence ICU outcome. Of the 1192 patients manifesting with no organ system failure (OSF) on the day of onset of SIRS, 140 (11.7%) died. Organ dysfunction evolved frequently in association with SIRS (63.4%). Non-survivors exhibited increased numbers of organ systems failed (1.7 ± 1.0 vs 0.7 ± 0.8, P < 0.001 [mean ± standard deviation]) and numbers of days of organ failure (0.87 ± 0.3 vs 0.5 ± 0.5 days, P < 0.001). Although individual organ system dysfunction had low specificity and sensitivity, the extent of OSF had significant impact on outcome (Table).

Table

Conclusions

Organ dysfunction occurs commonly with SIRS and in close temporal proximity with it; its duration and magnitude are strongly associated with increased mortality. Inclusion criteria employing the severity of organ dysfunction may help limit selection bias in sepsis trials.