Professional Documents
Culture Documents
60 5 3 < 1 (2–2·9)
1 Kyle RA, Gertz MA. Primary systemic amyloidosis: clinical and
40 p<0·0001 laboratory features in 474 cases. Semin Hematol 1995; 32: 45–59.
2 Apple FS, Wu AHB, Jaffe AS. European Society of Cardiology and
American College of Cardiology guidelines for redefinition of
20 myocardial infarction: how to use existing assays clinically and for
clinical trials. Am Heart J 2002; 144: 981–86.
3 Miller WL, Wright RS, McGregor CG, et al. Troponin levels in
0 patients with amyloid cardiomyopathy undergoing cardiac
transplantation. Am J Cardiol 2001; 88: 813–15.
4 Kyle RA, Greipp PR, O’Fallon WM. Primary systemic amyloidosis:
B Hazard ratio (95% CI) p multivariate analysis for prognostic factors in 168 cases. Blood 1986;
cTnI¶ 1·51 (1·15–1·98) 0·003 68: 220–24.
Ejection fraction§ 1·65 (1·25–2·18) 0·0004 5 Cueto-Garcia L, Reeder GS, Kyle RA, et al. Echocardiographic
Age‡ 1·26 (1·06–1·51) 0·01 findings in systemic amyloidosis: spectrum of cardiac involvement
24 h urine M protein† 1·35 (1·03–1·93) 0·03 and relation to survival. J Am Coll Cardiol 1985; 6: 737–43.
100 LV septal thickness|| 1·41 (1·03–1·78) 0·03
Divisions of Haematology and Internal Medicine (A Dispenzieri MD,
Score n MS (range months) R A Kyle MD, M A Gertz MD), Biostatistics (T M Therneau PHD),
80 1 44 16·5 (7–119)
Cardiovascular Diseases, and Internal Medicine (W L Miller MD,
2 178 11·1 (<1–158)
Survival (%)
39 2·2 (<1–37)
K Chandrasekaran MD, A S Jaffe MD), Laboratory Genetics
60 3
(J P McConnell PHD, A S Jaffe PhD), and Clinical Biochemistry and
Immunology (M F Burritt MD), Mayo Clinic, Rochester, MN 55905,
40 USA
p<0·0001
Correspondence to: Dr Angela Dispenzieri
20 (e-mail: dispenzieri.angela@mayo.edu)
0
0 20 40 60 80 100 120
Time (months)
CRIB II: an update of the clinical
cTnT (A) and cTnI (B) multivariate analysis of overall survival
in 261 patients with primary systemic amyloidosis risk index for babies score
MS=median survival. A: Continuous risk score was calculated as: R=3·19
√cTnT+0·52 √urine M-spike+0·025 age+0·34 ejection fraction; where Gareth Parry, Janet Tucker, William Tarnow-Mordi, for the UK
cTnT and urine M-spike were the actual values in g/L and g per 24 h, Neonatal Staffing Study Collaborative Group*
respectively; age was specified in years when less than 65, but for *See reference 2 for list of members
patients 65 years and older, age was taken as 65; ejection fraction was
one (<55%) or zero (55%). Overall risk score ranged from 0·77 to 5·2. The clinical risk index for babies (CRIB) score is a risk-
B: The optimum multivariate prediction model, designed independently adjustment instrument widely used in neonatal intensive care.
for cTnI was R=0·399 cTnI+0·497 ejection fraction+0·0247 age+0·338
septal thickness+0·373 √urine M-spike. The assignment of values for the Its appropriateness with contemporary data has been
variables is the same as for the cTnT model apart from: cTnI and septal questioned. We have examined these questions, developed a
thickness variables were each assigned a value of one (0·1 g/L, new five-item CRIB II score with data from a UK-wide sample of
>15 mm, respectively) or zero (<0·1 g/L, 15 mm, respectively). The infants admitted to neonatal intensive care in 1998–99, and
maximum value of this risk score was 3·18. *Square root of cTnT in
mg/L. †Square root of urine M protein in g per 24 h. ‡Age=number of shown how mortality after neonatal intensive care has fallen in
years when less than 65, but 65 for all those older 65 years. §Ejection the past 12 years. CRIB II provides a recalibrated and
fraction cut-off value of >55%. ¶cTnI <0·1 mg/L. ||Left-ventricular septal simplified scoring system that avoids the potential problems of
thickness as a cut-off value of >15 mm. early treatment bias. A valid and simple method of risk-
adjustment for neonatal intensive care is important to ensure
accurate assessment of quality of care. Such assessments
myocyte damage. Measurement of troponin values could should be done in tandem with national audit systems for
provide useful information for clinicians, and might neonatal intensive care, incorporating measures of morbidity as
potentially help to standardise risk comparisons. well as mortality.
Birthweight (g)
For personal use. Only reproduce with permission from The Lancet Publishing Group.
RESEARCH LETTERS
As an extension of work by Draper and co-workers (a grid Participating neonatal intensive care units
of probabilities of mortality by birthweight, gestation, and Airedale General Hospital, Keighley; Antrim Hospital, Antrim; Arrowe
Park Hospital, Upton; Birch Hill Hospital, Rochdale; Burnley General
sex)3 we developed and assessed CRIB II for infants of
Hospital, Burnley; City Hospital Birmingham, Birmingham; Derby City
32 weeks’ gestation, with further data available up to 1 h Hospital NHS Trust, Derby; Dewsbury District Hospital, Dewsbury;
after admission to neonatal intensive care, excluding FiO2. Erne Hospital, Enniskillen; Fairfield General Hospital, Bury;
To develop this instrument, we used data relating to a group Farnborough Hospital, Orpington; Gloucestershire Royal Hospital,
Gloucestershire; Good Hope Hospital, Sutton Coldfield; Greenwich
of 1886 infants consecutively admitted to 35 randomly
Healthcare Trust, London; Grimsby Maternity Hospital, Grimsby;
selected neonatal intensive care units contributing to the Hereford County Hospital, Hereford; Huddersfield Royal Infirmary,
UK Neonatal Staffing Study. Huddersfield; Jersey Maternity Hospital, St Helier; Kent and Canterbury
We fitted log-odds derived from the Draper Grid by use Hospital, Canterbury; Kettering General Hospital, Kettering; King’s Mill
Hospital, Sutton-In-Ashfield; Norfolk and Norwich Hospital, Norwich;
of logistic regression with mortality as the outcome. We
North Devon District Hospital, Barnstaple; Northampton General
then added other available variables relating to the period Hospital, Northampton; Northern General Hospital Sheffield, Sheffield;
up to 1 h after admission. Candidate variables identified by Northwick Park Hospital, Harrow; Nottingham University Hospital,
the UK Neonatal Staffing Study Steering Group were: Nottingham; Peterborough District Hospital, Peterborough; Prince
Charles Hospital, Merthyr Tydfil; Queen Elizabeth Hospital Gateshead,
mode of delivery, temperature at admission, surfactant
Gateshead; Rotherham District General Hospital, Rotherham; Royal
treatment, blood pH, SaO2, PaCO2, PaO2, and HCO3, and Alexandra Hospital, Brighton; Royal Devon and Exeter Hospital, Exeter;
diagnostic severity categories (established and validated by a Royal Hampshire County Hospital, Winchester; Royal Oldham Hospital,
panel of eight specialists in neonatal medicine). We used the Oldham; Royal Preston Hospital, Preston; Royal Surrey County Hospital,
Guilford; Royal United Hospital, Bath; Salisbury NHS Trust, Salisbury;
Akaike information criteria to identify the best model
Scunthorpe General Hospital, Scunthorpe; South Cleveland Hospital,
including birthweight, gestation, sex, admission Middlesbrough; South End Hospital, South End; Southern General
temperature, and base excess, and the coefficients of the Hospital, Glasgow; St George’s Hospital, London; St Mary’s Hospital,
final model were rounded to form CRIB II (figure). Portsmouth; St Paul’s Hospital, Cheltenham; St Peter’s Hospital,
Chertsey; St Thomas’ Hospital, London; Stoke Mandeville Hospital,
Application of CRIB II to the assessment group (1065
Aylesbury; The Ipswich Hospital NHS Trust, Ipswich; William Harvey
infants admitted to the remaining 19 units) gave a receiver- Hospital, Ashford; Wordsley Hospital, Stourbridge; Wycombe General
operating-characteristic area of 0·92 (SE 0·01). Both the Hospital, High Wycombe; York District Hospital, York.
Hosmer-Lemeshow goodness of fit test (2=4·30, 8 df,
p=0·829) and Cox’s assessment (slope p=0·642 and Contributors
constant p=0·141) showed adequate calibration. G Parry planned and undertook the analysis, and contributed to the
We had shown no difference in CRIB-adjusted mortality writing of the paper. J Tucker and W Tarnow-Mordi reviewed the analysis
for infants of 32 weeks’ gestation or less between large and plan and results, and contributed to the writing of the paper.
small neonatal intensive care units (odds ratio 1·19, 95% CI
0·70–2·02) with the UK Neonatal Staffing Study data. We Conflict of interest statement
repeated this test with CRIB II, and found that CRIB II- None declared.
adjusted mortality did not differ between large and small
units (odds ratio 1·06, 95% CI 0·70–1·60). Thus similar Acknowledgments
results were obtained with CRIB on the basis of information We thank all UK neonatal units and staff for their support. We thank
link audit nurses, nurse managers, and clinicians who obtained
up to 12 h after admission, and with CRIB II on the basis of prospective data on our behalf. We also wish to acknowledge the huge
information up to 1 h after admission. contributions of the late Douglas Richardson for his efforts in the
When we assessed CRIB II in relation to death or major promotion of risk adjustment and assessment of improvements in
cerebral abnormality (cystic leucomalacia or porencephalic neonatal intensive care worldwide. The views expressed in this paper are
those of the authors. This study was funded by the NHS R&D
cyst on cranial ultrasound, arising more than 10 days after Executive, Mother and Child Health Programme (grant number
birth) in the assessment group, CRIB II gave a receiver- MCH:6-7) and the Department of Health, UK. The funding source had
operating-characteristic area of 0·82 (SE 0·02). This area no role in study design, data collection, data analysis, data
was significantly higher than those of 0·79 (0·02) for CRIB, interpretation, or writing of the report.
0·80 (0·02) for gestation, 0·77 (0·02) for birthweight, and
0·80 (0·02) for the Draper grid.
1 The International Neonatal Network. The CRIB (clinical risk index
Available 1988–90 and 1993–94 CRIB algorithms were for babies) score: a tool for assessing initial neonatal risk and
poorly calibrated for data from 1998–99. CRIB II provides a comparing performance of neonatal intensive care units. Lancet
recalibrated and simplified scoring system, which avoids the 1993; 342: 193–98.
potential problems associated with use of FiO2 and data up to 2 UK Neonatal Staffing Study Group. Patient volume, staffing and
workload in relation to risk-adjusted outcomes in a random stratified
12 h after admission. Although this difference might be due sample of UK neonatal intensive care units: a prospective evaluation.
to participation of different units in the different time Lancet 2002; 359: 99–107
periods, our findings of falling hospital mortality since 1988 3 Draper ES, Manktelow B, Field D, James D. Prediction of survival
are similar to those of other studies.4 This fall could be for preterm births by weight and gestational age: retrospective
attributable to increasing clinical expertise, the introduction population based study. BMJ 1999; 319: 1093–97
4 Richardson DK, Gray JE, Gortmaker SL, Goldman DA, Pursley
of surfactant treatment from the late 1980s, and the growing DM. McCormick MC. Declining severity adjusted mortality:
use of antenatal steroids from the early 1990s. Future evidence of improving neonatal care. Pediatrics 1998; 102: 893–99
investigations should assess the effect on health outcomes 5 Halm EA, Chassin MR. Why do hospital death rates vary?
over time of improved clinical skills, increased uptake of N Engl J Med 2001; 345: 692–94
effective treatments, and improved service organisation.5
Before falling mortality can be judged a health gain, Medical Care Research Unit, School of Health and Related
information about the long-term health of neonatal intensive Research, University of Sheffield, Regent Court, 30 Regent Street,
Sheffield S1 4DA, UK (G Parry PhD); Dugland Baird Centre for
care survivors should be routinely assessed.4 Future
Research on Women’s Health, Department of Obstetrics and
investigations of neonatal intensive care need not only the
Gynaecology, University of Aberdeen, Aberdeen, UK (J Tucker PhD);
benchmark information presented here, but also appropriate
Westmead and Children’s Hospital at Westmead, University of
risk-adjustment instruments such as CRIB II, and valid Sydney, Sydney, Australia (W Tarnow-Mordi MRCP)
morbidity outcomes as measurement of quality and
performance. This is achievable through the introduction of Correspondence to: Dr Gareth Parry
a national neonatal intensive care audit system. (e-mail: g.parry@sheffield.ac.uk)