Background s/aims: In the US a low volume of admissions is associated with higher mortality in very low birth weight (VLBW) infants leading to recommendations for centralisation of neonatal services. We examined the likelihood of mortality, discharge, and transfer for VLBW and/or < 33+0 week gestation babies in relation to volume of admissions and care level in England.
Methods Competing risks regression, allowing clustering at the unit level, was used with data from the first episode of care. Units were classified by level of care and tertile of volume.
Results Following case-mix adjustment, relative to highest volume level three (highest intensity) units, top-tertile level one was associated with reduced risk of mortality (OR:0.05; CI:0.01–0.35; p=0.002). Level one and lower tertile level two units were associated with increased probability of discharge (eg top-tertile level one, OR:1.91; CI:1.43–2.55; p<0.001). Level one and level two units were less likely to transfer (eg top-tertile level one, OR:0.49; CI:0.33–0.73; p<0.001). These effects became statistically insignificant once ‘high risk’ babies (with congenital abnormalities, requiring surgery, and born < 29 weeks gestation) were removed.
Conclusions In this UK study we show reduced mortality in level one relative to level three units, and that this difference is explained by a less severe case-mix in lower level units. The majority of care for high risk babies in England is appropriately delivered by high-level units. In the US the case-mix of high- and low-level units is similar. We suggest a network based approach achieves the benefits of centralisation without the disadvantages.