Volume 5 Supplement 1

21st International Symposium on Intensive Care and Emergency Medicine

Open Access

Comparison of prediction of hospital mortality by ICU medical staff and referring parent team doctors

  • SJ Ramsay1,
  • CD Gomersall1,
  • GM Joynt1,
  • TA Buckley1 and
  • C Lam1
Critical Care20015(Suppl 1):P230

https://doi.org/10.1186/cc1297

Received: 15 January 2001

Published: 2 March 2001

Introduction

There is much interest in outcome prediction for ICU patients and many studies have evaluated the accuracy of predictions made by ICU medical staff [1]. There is little data on the accuracy of outcome prediction made by the ward-based doctors who refer these patients for ICU admission.

Method

As part of an on-going prospective study comparing the accuracy of prediction of hospital mortality by ICU medical staff and referring parent team doctors we analysed the first 100 completed data sets. The most senior doctor from both the referring parent team and receiving ICU team were asked to give their prediction of the likely hospital mortality for all emergency adult admissions to our 22 bedded ICU. Predictions as a percentage chance of dying in hospital were made at the time of referral to ICU, there was no conferring between the ICU and parent team and all results were confidential. MPM II0 was scored in all patients.

Results

Data was analysed on all patients who had either died in hospital or been discharged. The mean age was 59.6 years, 59% were males, and the overall mortality rate was 38%. The case mix was medical 55%, neurosurgical 18%, and other surgical specialties 28%. Data was ranked in order of predicted mortality and divided into equal deciles. Average predicted mortality (%) and observed mortality (%) were calculated for patients in each decile. Table 1 shows ICU doctors were better able to predict outcome over a wide range of mortalities compared to referring doctors (Table 2). The latter tended to under predict mortality except for the patients most at risk of death. The Hosmer-Lemeshow goodness of fit test showed a good fit for ICU doctors' prediction (?2 =10.24, P = 0.25), but not for referring doctors (?2 = 31.32, P < 0.001). MPM II0 predicted mortality poorly in this cohort (?2 =18.62, P = 0.02).
Table 1

ICU doctors' prediction

Decile

Av. predicted mortality (%)

Observed mortality (%)

1

0.0

10.0

2

2.8

10.0

3

9.5

10.0

4

11.0

10.0

5

20.0

50.0

6

28.0

30.0

7

42.0

50.0

8

54.0

60.0

9

76.0

70.0

10

92.0

80.0

Table 2

Referring doctors' prediction

Decile

Av. predicted mortality (%)

Observed mortality (%)

1

0.0

0.0

2

4.8

0.0

3

10.0

20.0

4

12.5

40.0

5

20.0

50.0

6

23.5

10.0

7

34.0

50.0

8

50.0

90.0

9

57.5

60.0

10

89.0

60.0

Conclusion

Doctors who refer patients to ICU for emergency admission are not able to predict accurately the hospital mortality of these patients and tend to underestimate mortality when compared to ICU doctors.

Authors’ Affiliations

(1)
Department of Anaesthesia, Intensive Care, The Chinese University of Hong Kong, Prince of Wales Hospital

References

  1. Cullen DJ, Chernow B: Predicting outcome in critically ill patients. Crit Care Med 1994, 22: 1345-1347.PubMedView ArticleGoogle Scholar

Copyright

© The Author(s) 2001

Advertisement