Introdução à Medicina II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

60
1 Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units Introdução à Medicina II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3 1. [email protected]; 2. [email protected]; 3. [email protected];

description

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units. Introdução à Medicina II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3 1. [email protected]; 2. [email protected]; 3. [email protected]; . - PowerPoint PPT Presentation

Transcript of Introdução à Medicina II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Page 1: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

1

Case-mix and Optimization of Mortality Prognostic

Scores in Portuguese Pediatric Intensive Care Units

Introdução à Medicina II

2011/2012Class 121

Advisors: Rosa Oliveira2 and Daniel Martins3

1. [email protected];

2. [email protected]; 3. [email protected];

Page 2: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Table of Contents

2Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

• Introduction• Contextualization

• State of the Art• Background

• Are scores discriminative? And calibrated?• Differences in calibration… Why?• Should scores be recalibrated?• Problems…

• Research Questions• Aims• Methods• Expected results• Results• References• Acknowledgments

Page 3: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

PICUs – Pediatric Intensive Care Units

Introduction

3Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

Introduction

Background

Research Questions

Aims

References

Acknowledgements

State of the Art

Contextualization

• A Pediatric Intensive Care Unit (PICU) is an area within a hospital specialized in the care of critically ill infants, children, and teenagers.

• The type and number of patients and the provided health care differs according to its location.

Frankel, Lorry R; DiCarlo, Joseph V (2003). “Pediatric Intensive Care”. In Bernstein, Daniel; Shelov, Steven P. Pediatrics for Medical Students (2nd ed.).

Methods

References

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

Methods

Page 4: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Introduction

Background

Research Questions

Aims

References

Acknowledgements

State of the Art

Methods

Consume human and financial resources

Provide adequate treatment for severely ill children

PICUs – Pediatric Intensive Care Units

Introduction

4PICUs – Pediatric Intensive Care Units

Portuguese PICUs

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

PICUs – Pediatric Intensive Care Units

Page 5: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Introduction

Background

Research Questions

Aims

References

Acknowledgements

State of the Art

Methods

PICUs – Pediatric Intensive Care Units

Introduction

5PICUs – Pediatric Intensive Care Units

Portuguese PICUs

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

Provide adequate treatment for severely ill children

Major health quality resource Pacient stratification

Evaluation

Mortality rate Assistance quality

PICUs – Pediatric Intensive Care Units

Page 6: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Introduction

Background

Research Questions

Aims

References

Acknowledgements

State of the Art

Methods

PICUs – Pediatric Intensive Care Units

Introduction

6PICUs – Pediatric Intensive Care Units

Portuguese PICUs

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

Provide adequate treatment for severely ill children

Major health quality resource Pacient stratification

Evaluation

Mortality rate Assistance quality

PICUs – Pediatric Intensive Care Units

Page 7: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Introduction

Background

Research Questions

Aims

References

Acknowledgements

State of the Art

Methods

PICUs – Pediatric Intensive Care Units

Introduction

7PICUs – Pediatric Intensive Care Units

Portuguese PICUs

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

Provide adequate treatment for severely ill children

Major health quality resource Pacient stratification

Evaluation

Mortality rate Assistance quality

PICUs – Pediatric Intensive Care Units

Page 8: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Introduction

Background

Research Questions

Aims

References

Acknowledgements

State of the Art

Methods

PICUs – Pediatric Intensive Care Units

Introduction

8PICUs – Pediatric Intensive Care Units

Portuguese PICUs

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

Provide adequate treatment for severely ill children

Major health quality resource Pacient stratification

Evaluation

Mortality rate Assistance quality

PICUs – Pediatric Intensive Care Units

Page 9: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Introduction

Background

Research Questions

Aims

References

Acknowledgements

State of the Art

Methods

PICUs – Pediatric Intensive Care Units

Introduction

9PICUs – Pediatric Intensive Care Units

Portuguese PICUs

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

Provide adequate treatment for severely ill children

Major health quality resource Pacient stratification

Evaluation

Mortalit rate Assistance quality

PICUs – Pediatric Intensive Care Units

Page 10: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Introduction

Background

Research Questions

Aims

References

Acknowledgements

State of the Art

Methods

Consume human and financial resources

PICUs – Pediatric Intensive Care Units

Introduction

10PICUs – Pediatric Intensive Care Units

Portuguese PICUs

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

Provide adequate treatment for severely ill children

PICUs – Pediatric Intensive Care Units

Page 11: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Mortality Prognostic Scores

PIM PRISM

Introduction

11PICUs – Pediatric Intensive Care Units

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

PICUs – Pediatric Intensive Care Units

PIM – Pediatric Index of Mortality; PRISM – Pediatric Risk of Mortality

Page 12: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Mortality Prognostic Scores

PIM PRISM

Introduction

12PICUs – Pediatric Intensive Care Units

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

PICUs – Pediatric Intensive Care Units

PIM – Pediatric Index of Mortality; PRISM – Pediatric Risk of Mortality

Page 13: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Introduction

13PICUs – Pediatric Intensive Care Units

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Expected results

Methods

Research Questions

Introduction

PICUs – Pediatric Intensive Care Units

PIM – Pediatric Index of Mortality; PRISM – Pediatric Risk of Mortality

PIM PRISM•Elective admission

•Subjacent conditions•Pupil reaction to light•Mecânical ventilation

•Blood Pressure•Blood alcalemia•Oxigen in blood

•Sistolic blood presure•Diastolic blood presure

•Cardic and respiratory frequency•Partial pressure of oxygen in the blood (PaO2)•Partial pressure of carbon dioxide in the blood

(PaCO2)•Glasgow Coma scale

•Pupil reactionsTP/TTP•Total bilirubin

•Potassium and calcium concentrations•Glucose index•Bicarbonate

• Observation during 1h •Observation during 24h

• Parameters in admission and one hour after •The hightest parameter in the first 24 hours

• Amount of information;• Duration of the observation period;

• Moment in which the information is collected. 1

1. Slater, A.M.B., FRACP, FJFICM; Shann, Frank MB, BS, MD, FRACP, FJFICM; the ANZICS Paediatric Study Group, The suitability of the Pediatric Index of Mortality (PIM), PIM2, the Pediatric Risk of Mortality (PRISM), and PRISM III for monitoring the quality of pediatric intensive care in Australia and New Zealand. Pediatric Critical Care Medicine, 2004. 5(5): p. 447-453.

Logit = (-4.873) + (values * Beta) + (0.021 * (absolute(SBP-120))) + (0.071 * (absolute base excess)) + (0.415 * (FiO2/PaO2))

Mortality Prediction Value= eLogit/ (1+eLogit)

Page 14: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

State of the Art

Carmen Ferández, J.L.-H., Jose C. Flores, Dolores Galaviz, Marta Rupérez, Kay B. Brandstrup, Amaya Bustinza, Prognosis in critically ill children requiring continuous renal replacement therapy. IPNA, 2005.

State of the Art

14

Background

Research Questions

Aims

References

Acknowledgements

Methods

Introduction

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

Page 15: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Are scores discriminative? And calibrated?

“Both PRISM and PIM scores have a good discriminatory performance. The

calibration with PRISM score is good but the PIM score displays poor

calibration.”

Taori RN, L. K., Tullu MS (2010). "Performance of PRISM (Pediatric Risk of Mortality) score and PIM (Pediatric Index of Mortality) score in a tertiary care pediatric ICU." Indian J Pediatr 77(3): 267-271.

Background

15

Research Questions

Aims

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

PIM – Pediatric Index of Mortality; PRISM – Pediatric Risk of Mortality

Page 16: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Diferences in calibration... Why?

Background

16

Research Questions

Aims

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

Different regions

Missing values

Recalibrations

→ Population case-mix→ Type of PICU→ Admission policy→ …

PICUs – Pediatric Intensive Care Units

Page 17: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Should scores be recalibrated?

“Subsequent validation of variables could improve the sensitivity and the

value of severity scoring in the future. Nevertheless, illness

severity scores will never be indicative of absolute irreversibility of disease or

impossibility of survival...” 2

“ In general, scoring systems should be used only in populations similar to

the reference population in which the prediction model was developed.” 1

Wells M, R.-F.J., Luyt DK, Dance M, Lipman J., Poor discriminatory performance of the Pediatric Risk of Mortality (PRISM) score in a South African intensive care unit. Crit Care Med., 1996. 24(9)

Unertl K, K.B., Prognostic scores in intensive care. Anaesthesist., 1997. 46(6): p. 471-80.

Background

17

Research Questions

Aims

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

Page 18: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Problems...

Background

18

Research Questions

Aims

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Research Questions

Introduction

→ Inclusion and exclusion criteria of the data-base;

→ Time of collection;

→ Human resources;

→ Algorithm descalibration through time; → …

Page 19: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

Recalibration Background

19

PIM

PRISM

PIM-2

PRISM III

Assessment and Optimization of Mortality Pronostic Scores in Portuguese Pediatric Intensive Care

“Are they doing a good job?”

MARTINS, D. A. A. 1; SANTOS, L. R.1; FIGUEIREDO, M. A. R. M.1; GUERRA, M. D. L.1; MAGALHÃES, R. S.1; FRANCISCO, R. M. H.1; REBELO, C. A. B. S. T.1; ABREU, C. A. M. L. M.1; LAIGINHAS, A. R. A.1; DIAS, M. O.1; CHAVES, J. G. A. C.1; OLIVEIRA, R. C. S.1,2;

Aim: To evaluate the Pediatric Risk of Mortality (…) and the Pediatric Index of Mortality (…) performance at Portuguese Pediatric Intensive Care Units, where PIM2-R is the recalibrated PIM2 model using the Australia and New Zealand Pediatric Intensive Care 2008 coefficients (which is for the first time applied and analyzed in Portuguese context).

Lack of calibration and discrimination in Portuguese reality

Research Questions

Aims

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

State of the Art

Background

Aims

Acknowledgements

Results

Methods

Background

Introduction

Research Questions

PIM-3

PIM – Pediatric Index of Mortality; PRISM – Pediatric Risk of Mortality

Page 20: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

20

References

Methods

State of the Art

Background

Research Questions

Aims

Expected results

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

Carmen Ferández, J.L.-H., Jose C. Flores, Dolores Galaviz, Marta Rupérez, Kay B. Brandstrup, Amaya Bustinza, Prognosis in critically ill children requiring continuous renal replacement therapy. IPNA, 2005.

PIM AND PRISM underestimated the risk of death in other realities

Underestimated risk of death in Portugal

Optimize of the quality of the health care provided

We look forward to...

Acknowledgements

Introduction

We expect...

PIM – Pediatric Index of Mortality; PRISM – Pediatric Risk of Mortality

Results

Page 21: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

• How do Pediatric Mortality Prognostic Scores (PIM-2) behave in Portuguese PICUs’ specific groups of diagnosis?

• Is it possible to enhance the performance of PIM-2 with a new recalibration?

• With what purpose are scores used in Portuguese PICUs’?

Research Questions

21

Research Questions

Aims

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Research Questions

Aims

Acknowledgements

Results

Methods

Background

Introduction

State of the Art

PICUs – Pediatric Intensive Care Units; PIM – Pediatric Index of Mortality

Page 22: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

• To assess calibration and discrimination of Pediatric Mortality Prognostic Scores in Portuguese PICUs and in the specific diagnosis groups.

• To recalibrate PIM-2, in the broad-spectrum patients and to specific diagnosis group.

• To evaluate the new model concerning calibration and discrimination

• To understand clinicians awareness of the purpose and use of prognostic mortality scores in Portuguese PICUs.

Specific aims

22

Aims

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Research Questions

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

PICUs – Pediatric Intensive Care Units; PIM – Pediatric Index of Mortality

Page 23: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

23

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Research Questions

Aims

Target Population

Methods

23Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

The target population will be all children admitted in Portuguese PICUs.

PICUs – Pediatric Intensive Care Units

Page 24: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Research Questions

Aims

Sample

Methods

24Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

Sources: PICUs from Hospital de Santa Maria, Hospital de D. Estefânia, Hospital Fernando da Fonseca, Hospital Geral de S. João, Hospital Pediátrico de Coimbra, Hospital Garcia de Orta, Hospital Geral de Santo António, Hospital da Misericórdia de Paredes and Hospital de Faro.

Children were included in different groups of diagnosis.

PICUs – Pediatric Intensive Care Units

Page 25: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Research Questions

Aims

Sample

Methods

25Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

Age(mean) : 2005-2006: 6 y 1 m; 2007-2008: 6 y 1 m.

Lenght of stay(mean): 2005-2006: 182h; 2007-2008: 127h.

43%

57%

Admission Period2005/2006 2007/2008

Male Female

55.8 56.356.343.7

Gender

Planed Unplaned

36.5

63.5

39.6 60,4

Admission Type

Page 26: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Research Questions

Aims

Sample

Methods

26Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

Accidents/

External C

auses

Cardiovascu

lar

Neurological

Respira

toryRenal

Gastrointesti

nal

Oncological

Metabolic/E

ndocrinal/E

lectrolyti

c

Post-proce

dure

Post-operatory

Others

16

2.1

9.5

22.8

0.6 0.3 1.5 2.4 2.7

32.3

9.8

17.8

1.1

9.7

18

0.7 0.9 0.4 1.3 0.2

41.1

8.8

Groups Diagnosis

2005/20062007/2008

Page 27: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Research Questions

Aims

Validation

Calibration Discrimination

Methods

27

Standardized Mortality Ratio (SMR)

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

Page 28: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Research Questions

Aims

Validation

Calibration Discrimination

Methods

28

Standardized Mortality Ratio (SMR)

Observed deaths / expected deaths

SMR<1.00 SMR=1 SMR>1.00“we are

doing better than

expected”

“we are doing as

expected”

“we are doing worse than expected”

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

Page 29: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Research Questions

Aims

Validation

Calibration Discrimination

Methods

29Case-mix and Optimization in Mortality Prognostic Scores in Portuguese PICUs

Standardized Mortality Ratio (SMR)

Hosmer-Lemeshow Test

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

Page 30: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Research Questions

Aims

Validation

Calibration Discrimination

Methods

30

Standardized Mortality Ratio (SMR)

Relative Operating Characteristic (ROC)

Curves

Sensitivity vs. 1-Specificity for every

value of the score

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

Page 31: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

References

Acknowledgements

Methods

Introduction

State of the Art

Background

Research Questions

Aims

Methods

31Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

Validation

Calibration DiscriminationStandardized Mortality Ratio

Sensitivity vs. 1-Specificity for every

value of the score

Area under the (AUC)ROC Curve

Relative Operating Characteristic (ROC)

Curves

Page 32: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

32

References

Acknowledgements

Methods

State of the Art

Background

Research Questions

Aims

Methods

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

AUC – Area Under the ROC Curve

0,70≤AUC<0,80 0,80≤AUC<0,90 0,90≤AUC<1,00 AUC=1,00Acceptable Good Execlent Perfect

Page 33: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

33

Methods

Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

Main study

Surveys of use

Grounding of conclusions

https://docs.google.com/spreadsheet/viewform?pli=1&formkey=dHZpbU9hX1h6cFEyWExnTzNzRHplcnc6MQ#gid=0

Page 34: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

34

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

ResultsAgreement* (%)

Clinicians… 1 2 3 4 5 NAwereness of Mortality Prognostic Scores :- Concept 0 0 16,7 55,6 2,78 18

- Use in Portuguese PICUs 0 0 10,5 21,1 47,4 19

- Use in the their specific PICU 0 10,5 21,1 47,4 21,1 19

Access to the scores 0 26,3 10,5 47,4 15,8 19

Idea of scores Calibration to Portuguese reality 5,3 10,5 42,1 36,8 5,3 19

Idea of scores usefulness in Portuguese PICUs 0 0 5,3 89,5 5,3 19

*1 - Total disagreement, …, 5 - Total agreement

Page 35: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

35

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

ResultsAgreement* (%)

Clinicians… 1 2 3 4 5 NAwereness of Mortality Prognostic Scores :- Concept 0 0 16,7 55,6 2,78 18

- Use in Portuguese PICUs 0 0 10,5 21,1 47,4 19

- Use in the their specific PICU 0 10,5 21,1 47,4 21,1 19

Access to the scores 0 26,3 10,5 47,4 15,8 19

Idea of scores Calibration to Portuguese reality 5,3 10,5 42,1 36,8 5,3 19

Idea of scores usefulness in Portuguese PICUs 0 0 5,3 89,5 5,3 19

*1 - Total disagreement, …, 5 - Total agreement

Page 36: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

36

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

ResultsAgreement* (%)

Clinicians opinion … 1 2 3 4 5 NImpact of:the score use in the their specific PICU 0 0 15,8 78,9 5,3 19

recommended score (in their decision) 57,9 36,8 0 5,3 0 19

a recalibrated score (in their decision) 10,5 26,3 21,1 36,8 5,3 19scores usefulness in clinical practice or PICU structure 5,6 38,9 44,4 11,2 0 18

Of scores abillity todiscriminate the quality of the health care 0 0 21,1 68,4 10,5 19

provide patient stratification 0 0 36,8 57,3 5,3 19

*1 - Total disagreement, …, 5 - Total agreement

Page 37: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

37

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

ResultsAgreement* (%)

Clinicians opinion … 1 2 3 4 5 NImpact of:the score use in the their specific PICU 0 0 15,8 78,9 5,3 19

recommended score (in their decision) 57,9 36,8 0 5,3 0 19

a recalibrated score (in their decision) 10,5 26,3 21,1 36,8 5,3 19scores usefulness in clinical practice or PICU structure 5,6 38,9 44,4 11,2 0 18

Of scores abillity todiscriminate the quality of the health care 0 0 21,1 68,4 10,5 19

provide patient stratification 0 0 36,8 57,3 5,3 19

*1 - Total disagreement, …, 5 - Total agreement

Page 38: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

38

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Results

Clinicians … Yes (%) No(%) NKnowledge of a model 94,7 5,3 19Opinion on the necessity of recalibration in each country 57,9 42,1 19

Page 39: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

39

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Results

Clinicians … Yes (%) No(%) NKnowledge of a model 94,7 5,3 19Opinion on the necessity of recalibration in each country 57,9 42,1 19

Page 40: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

40

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

Results

References

PIM-2 AUC

2002 0,639

2006 0,644

2008 0,647

2005/2006

AUC – Area Under the Curve; PIM – Pediatric Index of Mortality

1-Specificity

Sen

sitiv

ity

Page 41: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

41

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Results

PIM-2 AUC

2002 0,633

2006 0,632

2008 0,645

2007/2008

AUC – Area Under the Curve; PIM – Pediatric Index of Mortality

1-Specificity

Sens

itivi

ty

Page 42: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

42

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Results

Page 43: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

43

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Results

Page 44: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

44

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Results

Page 45: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

45

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Results

AUC

PIM2

2002 2006 2008

2005/2006

2007/2008

2005/2006

2007/2008

2005/2006

2007/2008

Accidents/External Causes

0,422 0,498 0,432 0,525 0,405 0,506

Cardiovascular 0,917 0,667 0,833 0,667 0,833 0,667

Neurologic 0,603 0,399 0,609 0,413 0,596 0,406

Respiratory 0,661 0,688 0,657 0,695 0,665 0,693

Postoperative 0,770 0,747 0,755 0,747 0,758 0,699

Others 0,657 0,745 0,674 0,725 0,661 0,716

AUC – Area Under the Curve; PIM – Pediatric Index of Mortality

Page 46: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

46

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Results

AUC

PIM2

2002 2006 2008

2005/2006

2007/2008

2005/2006

2007/2008

2005/2006

2007/2008

Accidents/External Causes

0,422 0,498 0,432 0,525 0,405 0,506

Cardiovascular 0,917 0,667 0,833 0,667 0,833 0,667

Neurologic 0,603 0,399 0,609 0,413 0,596 0,406

Respiratory 0,661 0,688 0,657 0,695 0,665 0,693

Postoperative 0,770 0,747 0,755 0,747 0,758 0,699

Others 0,657 0,745 0,674 0,725 0,661 0,716

AUC – Area Under the Curve; PIM – Pediatric Index of Mortality

Page 47: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

47

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Results

AUC

PIM2

2002 2006 2008

2005/2006

2007/2008

2005/2006

2007/2008

2005/2006

2007/2008

Accidents/External Causes

0,422 0,498 0,432 0,525 0,405 0,506

Cardiovascular 0,917 0,667 0,833 0,667 0,833 0,667

Neurologic 0,603 0,399 0,609 0,413 0,596 0,406

Respiratory 0,661 0,688 0,657 0,695 0,665 0,693

Postoperative 0,770 0,747 0,755 0,747 0,758 0,699

Others 0,657 0,745 0,674 0,725 0,661 0,716

AUC – Area Under the Curve; PIM – Pediatric Index of Mortality

Page 48: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

48

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Results

• New Algorithm:

o PIM-2 IM=(-0.015 x SBP)+(4.270 x PUPIL REACTION)+ - 0.011 x mm)+(0.040 x BASE EXCESS)+(- 0.695 x VENTILATION)+(- 3.019 x PLANED ADMISSION)+(-1.099 x POST OPERATORY)+(19.805 x CARDIAC SURGERY)+(0.702 x RECOVERY)+(0.426 x DIGESTIVE) + 2.779

Page 49: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

49

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Results

PIM-2 AUC

2002 0,633

2006 0,632

2008 0,645

2007/2008

AUC – Area Under the Curve; PIM – Pediatric Index of Mortality

1-Specificity

Sens

itivi

ty

PIM-2 AUC

IM 0,875

Page 50: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

50

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Results

Page 51: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

51

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

ResultsGroup AUC Chi-square p-value

Accidents/External Causes

0,848 19,336 0,013

Cardiovascular 0,500 5,000 0,172

Neurologic 0,865 7,326 0,502

Respiratory 0,674 10,329 0,243

Postoperative 0,731 7,816 0,452

Others 0,745 8,787 0,361

Group AUC Chi-square p-value

Accidents/External Causes

0,525 11,043 0,199

Cardiovascular 0,667 4,841 0,184

Neurologic 0,413 7,469 0,487

Respiratory 0,695 9,219 0,324

Postoperative 0,747 8,815 0,358

Others 0,725 12,430 0,133

PIM-2 IM

PIM-2 2006

Page 52: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

52

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

ResultsGroup AUC Chi-square p-value

Accidents/External Causes

0,848 19,336 0,013

Cardiovascular 0,500 5,000 0,172

Neurologic 0,865 7,326 0,502

Respiratory 0,674 10,329 0,243

Postoperative 0,731 7,816 0,452

Others 0,745 8,787 0,361

Group AUC Chi-square p-value

Accidents/External Causes

0,525 11,043 0,199

Cardiovascular 0,667 4,841 0,184

Neurologic 0,413 7,469 0,487

Respiratory 0,695 9,219 0,324

Postoperative 0,747 8,815 0,358

Others 0,725 12,430 0,133

PIM-2 IM

PIM-2 2006

Page 53: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

53

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Conclusions and Discussion

PIM-2 showed acceptable discrimination but poor calibration in specific diagnosis groups and in general Portuguese PICUs.

Recalibrated version of PIM-2 showed better discrimination and calibration either in specific diagnosis groups either general Portuguese PICUs than the ones used nowadays.

Mortality Prognostic Scores’ importance is recognized in Portuguese PICUs, however:

Clinicians don’t feel the need of a recalibration Scores don’t frequently lead to a change in clinical practice.

Page 54: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

54

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Conclusions and Discussion

PIM-2 showed acceptable discrimination but poor calibration in specific diagnosis groups and in general Portuguese PICUs.

Recalibrated version of PIM-2 showed better discrimination and calibration either in specific diagnosis groups either general Portuguese PICUs than the ones used nowadays.

Mortality Prognostic Scores’ importance is recognized in Portuguese PICUs, however:

Clinicians don’t feel the need of a recalibration Scores don’t frequently lead to a change in clinical

practice.

Page 55: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

55

Results

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Introduction

References

Conclusions and Discussion

PIM-2 showed acceptable discrimination but poor calibration in specific diagnosis groups, and in general Portuguese PICUs.

Recalibrated version of PIM-2 showed better discrimination and calibration either in specific diagnosis groups and general Portuguese PICUs than the ones used nowadays.

Mortality Prognostic Scores’ importance is recognized in Portuguese PICUs, however:

Clinicians don’t feel the need of a recalibration Scores don’t frequently lead to a change in clinical

practice.

Page 56: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

• Anthony R. Brady, D.H., Stephanie Black, Sam Jones, Kathy Rowan, Gale Pearson, Jane Ratcliffe and Gareth J. Parry, Assessment and Optimization of Mortality Prediction Tools for Admissions to Pediatric Intensive Care in the United Kingdom. Pediatrics, 2006. 117;

• Carmen Ferández, J.L.-H., Jose C. Flores, Dolores Galaviz, Marta Rupérez, Kay B. Brandstrup, Amaya Bustinza, Prognosis in critically ill children requiring continuous renal replacement therapy. IPNA, 2005;

• César Sánchez, J.L.-H., Santiago Mencía, Javier Urbano, Angel Carrillo and José María Bellón, Clinical severity scores do not predict tolerance to enteral nutrition in critically ill children. British Journal of Nutrition, 2009. 102: p. 191–194;

• Esra Arun Ozer, A.K., Berrak Sarioglu, Oya Halicioglu, Sumer Sutcuoglu, and Isin Yaprak, The Comparison of PRISM and PIM Scoring Systems for Mortality Risk in Infantile Intensive Care. Journal of Tropical Pediatrics, 2004. 50;

References

56Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Results

Introduction

Page 57: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

• Ganapathy Balakrishnan, T.A., David Hallworth, Neil S Morton, Prospective evaluation of the Paediatric Risk of Mortality (PRISM) score. Archives ofDisease in Childhood, 1992. 67: p. 196-200.

• Okay, G.A.C.A.F.D.A.F.T.S., Application of the pediatric risk of mortality (PRISM) score and determination of mortality risk factors in a tertiary pediatric intensive care unit. Clinics, 2010. 65.

• Pollack MM, R.U., Getson PR, Pediatric risk of mortality (PRISM) score. Crit Care Med, 1998. 16(11): p. 1110-6.

• Slater, A.M.B., FRACP, FJFICM; Shann, Frank MB, BS, MD, FRACP, FJFICM; the ANZICS Paediatric Study Group, The suitability of the Pediatric Index of Mortality (PIM), PIM2, the Pediatric Risk of Mortality (PRISM), and PRISM III for monitoring the quality of pediatric intensive care in Australia and New Zealand. Pediatric Critical Care Medicine, 2004. 5(5): p. 447-453.

• S. Prieto Espuñes, J.L.-H.C., C. Rey Galán, A. Medina Villanueva, A. Concha Torre y P. Martínez Camblor, Índices pronósticos de mortalidad en cuidados intensivos pediátricos. An Pediatr (Barc), 2007. 66: p. 345-50.

References

57Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Expected results

Introduction

Page 58: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

• S M Tibby, D.T., M Festa, S Hanna, M Hatherill, G Jones, P Habibi, A Durward, I A Murdoch, A comparison of three scoring systems for mortality risk among retrieved intensive care patients. Arch Dis Child, 2002. 87: p. 421-425;

• Vanessa Feller Martha, P.C.R.G., Jefferson Pedro Piva, Paulo Roberto Einloft, Francisco Bruno, Viviane Rampon, Comparação entre dois escores de prognóstico (PRISM e PIM) em unidade de terapia intensiva pediátrica. Jornal de Pediatria, 2005. 81(3): p. 259-64.

References

58Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

References

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Expected results

Introduction

Page 59: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

We gratefully acknowledge :• Professor Altamiro Costa Pereira;• Phd Armando Teixeira-Pinto;• Msc Rosa Oliveira;• Daniel Martins;

for their guidance and time

• And you;

For your attention.

PICUs – Pediatric Intensive Care Units

Acknowledgments

59Case-mix and Optimization of Mortality Prognostic Scores in Portuguese Pediatric Intensive Care Units

Methods

State of the Art

Background

Research Questions

Aims

Acknowledgements

Expected results

Introduction

References

Page 60: Introdução à Medicina  II 2011/2012 Class 12 1 Advisors: Rosa Oliveira 2 and Daniel Martins 3

The End

60

Introdução à Medicina II

2011/2012Class 121

Advisors: Rosa Oliveira2 and Daniel Martins3

1. [email protected];

2. [email protected]; 3. [email protected];

Questions?