Elsevier

The Lancet

Volume 359, Issue 9301, 12 January 2002, Pages 99-107
The Lancet

Articles
Patient volume, staffing, and workload in relation to risk-adjusted outcomes in a random stratified sample of UK neonatal intensive care units: a prospective evaluation

https://doi.org/10.1016/S0140-6736(02)07366-XGet rights and content

Summary

Background

UK recommendations suggest that large neonatal intensive-care units (NICUs) have better outcomes than small units, although this suggestion remains unproven. We assessed whether patient volume, staffing levels, and workload are associated with risk-adjusted outcomes, and with costs or staff wellbeing.

Methods

186 UK NICUs were stratified according to volume of patients, nursing provision, and neonatal consultant provision. Primary outcomes were hospital mortality, mortality or cerebral damage, and nosocomial bacteraemia. We studied 13 515 infants of all birthweights consecutively admitted to 54 randomly selected NICUs. Multiple logistic regression analyses were done with every primary outcome as the dependent variable. Staff wellbeing and stress were assessed by anonymous mental health index (MHI)-5 questionnaires.

Findings

Data were available for 13 334 (99%) infants. High-volume NICUs treated the sickest infants and had highest crude mortality. Risk-adjusted mortality and mortality or cerebral damage were unrelated to patient volume or staffing provision; however, nosocomial bacteraemia was less frequent in NICUs with low neonatal consultant provision (odds ratio 0·65, 95% CI 0·43–0·98). Mortality was raised with increasing workload in all types of NICUs. Infants admitted at full capacity versus half capacity were about 50% more likely to die, but there was wide uncertainty around this estimate. Most staff had MHI-5 scores that suggested good mental health.

Interpretation

The implications of this report for staffing policy, medicolegal risk management, and ethical practice remain to be tested. Centralisation of only the sickest infants could improve efficiency, provided that this does not create excessive workload for staff. Assessment of increased staffing levels that are closer to those in adult intensive care might be appropriate.

Introduction

One to two infants per 100 livebirths need at least 1 h of neonatal intensive care,1 but data about sustained neonatal intensive care are not routinely available because definitions vary.2 Low-volume, high-cost specialties such as neonatal intensive care aim to provide specialist care of the highest quality as efficiently as possible, and to maintain access.3 Since publication of the Short Report in 1980,4 policy documents have consistently recommended a tiered neonatal care system, and lend support to the view that high patient throughput or volume would raise accrued clinical and nursing expertise and therefore achieve the best outcomes.3

The definition of intensive care changes with evolving technology, but in general, neonatal care is classified as units that provide: no intensive care (special-care baby units); limited intensive care, but that transfer complex problems; full intensive care for infants who are born at the unit, but that accept no transfers (non-trading neonatal intensive-care units [NICUs]); and full intensive care for infants who are born in the unit or who are transferred in utero or postnatally.

In the first phase of this study,5 we showed wide variation in yearly levels of neonatal intensive-care activity, staffing, and skillmix in the UK. However, UK standards recommend that NICUs should have specific staffing levels and be designated as an NICU only if they attain at least 500 intensive-care days per year. This latter recommendation implied further centralisation of UK neonatal intensive-care services.2 However, authors of this document also noted that empirical evidence was needed to lend support to these recommendations.

Evidence of a relation between NICU performance and patient volume (and by implication, accrued expertise) of neonatal intensive care is not consistent in the UK. Data from reports comparing tertiary (and high volume) and non-tertiary centres in both Scotland6 and Trent7 in 1987–90 suggested improved survival in tertiary centres. Authors of a Californian cohort study8 reported improved outcomes for a high volume of neonatal intensive care in large tertiary NICUs. In population-based comparative studies (between the UK and Australia) of both neonatal intensive care9 and paediatric intensive care,10 there was lower adjusted neonatal and paediatric mortality in Australia than in the UK. The authors of these studies suggested that performance of NICUs and paediatric intensive-care units (PICUs) might be better in Australia partly because units are more than twice the size of those in the UK, with higher levels of staffing and specialisation and more centralised models of care provision. Conversely, population-based UK studies11, 12 have detected no differences in risk-adjusted mortality by unit size, and a cohort study in the UK and Republic of Ireland13 has shown no evidence of a volume effect on survival of very preterm infants (<26 weeks).

Findings of studies in adult intensive-care units tend to show no association between overall patient-volume measures and adjusted mortality,14, 15 although Goldfrad and Rowan15 report improved risk-adjusted mortality associated with higher admission rates per week in non-surgical patients than in those having surgery. Results of studies of association between volume and outcome in paediatric intensive care10, 16 seem to show a closely similar inconsistency. However, adult intensive-care studies in the UK have consistently shown relations between high workload measures and deteriorating performance.17, 18

We have refined the design of a study in the USA of paediatric intensive care,16 which showed an inverse relation between risk-adjusted survival and patient volume, but improved survival in hospitals with a paediatric intensivist. We (UK Neonatal Staffing Study, UKNSS) investigated prospectively the relation between UK norm-referenced categories of patient volume, staffing levels, and workload and hospital mortality, mortality or major cerebral abnormality, and probable nosocomial bacteraemia.19

Section snippets

Study design

To identify the groups of units required for the factorial design, a national census was done. Phase I census data5 identified 186 UK NICUs (units intending to provide sustained neonatal intensive care). The census was done without nationally agreed and verifiable definitions of neonatal intensive care, nurse or cot establishment, or unit size.20 Yearly data about unit activity and staffing provision allowed stratification into 12 types of NICU with a three (high, medium, and low volume) by two

Results

Of 14611 infants consecutively admitted to 54 NICUs, information was available for 14343 infants (98%, table 1). Of these, 13515 were eligible admissions, with complete data for 13334 (99%). There were 393 deaths in hospital (3%), and just over four-fifths of these were attributable to the participating hospitals of care (table 1). The 65 non-attributable deaths were because of lethal congenital conditions or complex cardiac surgery. Mean birthweight ranged from 2266 g (SD 986) in unit type one

Discussion

Our study showed that high-volume NICUs in the UK cared for sicker infants, were busier, and had higher crude mortality and morbidity than medium-volume and low-volume units. After adjustment for initial clinical risk and illness severity, the performance of high-volume NICUs was similar to that of medium-volume and low-volume NICUs. Low-volume NICUs transferred more infants in the first 24 h than medium-volume or high-volume units, and these were sick infants destined for secondary or tertiary

References (39)

  • Report of the Working Group on Neonatal Intensive Care to CSAG

    Neonatal intensive care: access to and availability of specialist services

    (1995)
  • Perinatal and neonatal mortality: second report

    (1980)
  • J Tucker et al.

    UK Neonatal Intensive Care Services in 1996

    Arch Dis Child Fetal Neonatal Ed

    (1999)
  • FieldD et al.

    Survival and place of treatment after premature delivery

    Arch Dis Child Fetal Neonatal Ed

    (1990)
  • PhibbsCS et al.

    The effects of patient volume and level of care at the hospital of birth on neonatal mortality

    JAMA

    (1996)
  • Risk adjusted and population-based studies of outcome for high risk infants in Scotland and Australia

    Arch Dis Child Fetal Neonatal Ed

    (2000)
  • RHB De Courcy Wheeler et al.

    Use of the CRIB (clinical risk index for babies) score in prediction of neonatal mortality and morbidity

    Arch Dis Child Fetal Neonatal Ed

    (1995)
  • FieldD et al.

    Survival and place of delivery following preterm birth: 1994–96

    Arch Dis Child Fetal Neonatal Ed

    (1999)
  • CosteloeK et al.

    The EPICure study: outcomes to discharge from hospital for infants born at the threshold of viability

    Paediatrics

    (2000)
  • Cited by (240)

    View all citing articles on Scopus

    Members listed at end of report

    View full text