Prompt admission to critical care for deteriorating ward patients

Clinical data used for analysis of the (SPOT)light study<br><b>Abstract</b><br>Background<br>Where critical care provision is limited, some ward patients may require admission when the critical care unit is full. We used this service constraint to explore the effect of prompt admission to critical care on survival.<br>Methods and Findings<br>This was a prospective cohort study of consecutive deteriorating ward patients assessed for critical care in 49 NHS hospitals (1 November 2010 — 31 December 2011). Delay in admission to critical care was timed from the bedside assessment. The primary endpoint was 90-day mortality. Cases were admitted within four hours. Controls included delayed admissions (4-168 hours), and those not admitted. A comparison of mortality between these groups would be biased if those admitted promptly were more acutely ill. To account for this bias, we used the effect of occupancy on prompt admission in a bivariate probit instrumental variable (IV) analysis. To isolate the prompt versus delayed comparison, we repeated the analysis in the subgroup initially offered critical care.<br>12495 patients from 48 hospitals were available for analysis of which 3279 (26%) were initially offered critical care. 2492 (20%) patients were admitted promptly, and 2002 controls (16%) were delayed admissions. As expected, prompt admissions were sicker (by 4·4 ICNARC physiology points [95%CI 4·1-4·8] on average) with a higher unadjusted mortality (37·4% versus 28·7%). Median time to admission was 3 hours (IQR 1-9) rising to 12 hours (IQR 5-29) when critical care was refused at the initial assessment. The unit was fully occupied at the time of 996 (8%) ward assessments. These assessments were less likely to lead to offers of critical care (p&lt;0·0001), or prompt admission (p&lt;0·0001), and more likely to be followed by death on the ward (p=0·0001). Overall, there were 3797 (30·4%) deaths by 90 days. Risk adjusted outcomes were equivalent (HR 1·00 [95%CI 0·88-1·13] p=0·988), but using the IV model to avoid unmeasured confoundering, prompt admission reduced absolute mortality by 10·0% (95%CI 0·4%-19·8%, p= 0·043). In the subgroup offered critical care, the reduction in mortality was not significant (16·7%, 95%CI -0·9% to 34·0%, p= 0·063).<br>Conclusion<br>Prompt admission to critical care reduces 90-day mortality when compared to controls who are either admitted late, or never admitted.<br>