Introduction Recent studies used present cut-off levels (n>50 VLBW-infants/year, e.g.) for NICU intergroup comparisons and identified a soft volume-outcome relationship, explaining ∼10% variation of VLBW-mortality. Statistics couldn’t identify an appropriate volume-based cut-off for unit shut-downs in order to optimize overall outcome.

This failure is due to the fact that – even between units of comparable size – mortality rates vary by a factor of 3, raising the hypothesis that volume-driven NICU shut-downs have minor beneficial effects on overall outcome. The aim of this study was to investigate changes in mortality rates when NICU closings are based on either volume size (VLBW/year) or outcome quality (mortality).

Methods NICU-volumes and mortality rates given in recent publications (Vermont-Oxford-Network, n = 22446 VLBW-infants, and NRW-neonatal-register, Germany) were used.

Centers were stepwise excluded using either volume load or outcome quality. Apparent overall mortality rate was calculated after randomly allocating infants from closed down units to one of the remaining centers. For each exclusion step, this procedure was repeated 1000 times. A 10% effect on mortality with volume size was assumed.

Result Quality-based strategies are considerably more effective in improving VLBW-outcome. Volume based strategy needs to close down more than 50% of the units (representing 25% of VLBW-infants) to achieve more than 5%; its maximum effect is only 20%.

Discussion This study clearly shows that volume-based cut-offs alone are not appropriate to decrease VLBW mortality, but should be combined with cut-offs using quality indicators.