Transcript Pediatric Trauma Assessment and Management Database
P EDIATRIC T RAUMA A SSESSMENT AND M ANAGEMENT DATABASE A TRAUMA REGISTRY-VPS PARTNERSHIP
Katherine T. Flynn-O’Brien, MD Mary E. Fallat, MD Tom B. Rice, MD Christine M. Gall, RN, MS, DrPH Frederick P. Rivara, MD
VPS User Conference| March 24-26, 2015
Outline
Motivation What we did How we did it I.
What we discovered II.
III.
More data (and better data) Risk-adjustment modeling Processes of care Brainstorming
Motivation
Limited ability to study pediatric trauma NTDB / Pediatric TQIP Virtual Pediatric ICU Systems (VPS) UDSMR, HCUP, PHIS, MarketScan
Objective
Create a comprehensive pediatric trauma database to assess quality of care in critically injured children utilizing minimal new resources.
Methods
Merged 3 databases Trauma Registry (TR) Virtual PICU Systems (VPS) data PTAM-specific RedCap 5 Level I/II PTC All children discharged from PICU CY 2013
Process
Trauma Registry
(local export)
VPS
(central export)
PTAM Additional data elements
(data entry)
95.5% match
Additional variables
All CT scans ICPM placement Mech. VTE proph.
Lab upon arrival Initiation of feeds Bowel regimen C-spine clearance Hgb prior to transfusion FAST Alcohol screening & counseling (TQIP variables)
I. More data | Better data
Breadth & depth
Care continuum
Pre-hospital ED PICU You are here.
Floor Discharge
Care continuum
• Vitals • GCS • Transfer
Pre hospital ED arrival
• Vitals • GCS • Labs* • Vitals • Labs • Vent data
ICU stay Floor
• Nutrition • Constipation • VTE ppx • Disposition • LOS
Discharge
Care continuum
Variable
GCS Pulse Blood Pressure Hemoglobin Base Deficit AST ALT Hypoxemia PT/PTT CT scans
Pre-hospital X X X X ED X X X X X X X XX X X PICU X X X X X XX XX XX X X Floor (X) X
Care continuum
Variable
GCS Pulse Blood Pressure Hemoglobin Base Deficit AST ALT Hypoxemia PT/PTT CT scans
Pre-hospital X X X X ED X X X X X X X XX X X PICU X X X X X XX XX XX X X Floor (X) X
Care continuum
Variable
GCS Pulse Blood Pressure Hemoglobin Base Deficit AST ALT Hypoxemia PT/PTT CT scans
Pre-hospital X X X X ED X X X X X X X XX X X PICU X X X X X XX XX XX X X Floor (X) X
Care Continuum
What is the child’s cognitive/physiologic status immediately after injury? What resuscitation is, or is not, occurring prior to ICU arrival?
How may this information change management in the ICU?
More data
Better data
Complications Cardiac arrest CLABSI Unplanned return to the ICU Pneumonia Re-intubation
Better data
Complications Cardiac arrest CLABSI Unplanned return to the ICU Pneumonia Re-intubation
4 37
Better data
Complications Cardiac arrest CLABSI Unplanned return to the ICU Pneumonia Re-intubation
1 2
Better data
Complications Cardiac arrest CLABSI Unplanned return to the ICU Pneumonia Re-intubation
3 14
Better data
Complications Cardiac arrest CLABSI Unplanned return to the ICU Pneumonia Re-intubation
5 5
Better data
Complications Cardiac arrest CLABSI Unplanned return to the ICU Pneumonia Re-intubation
0 20
Better data
Comorbidities Hx of CVA Prematurity Respiratory distress
Better data
Comorbidities Hx of CVA Prematurity Respiratory distress
1 1
Better data
Comorbidities Hx of CVA Prematurity Respiratory distress
6 18
Better data
Comorbidities Hx of CVA Prematurity Respiratory distress
4 14
II. Risk adjustment modeling
Mortality
Trauma Registry
Model building Model diagnostics Multiple imputation PIM2 PRISM3 PELOD
VPS
Mortality
Model AUC R 2 value AIC TR-only VPS-only
0.9360
0.9917
0.5286
0.6723
127.57
95.84
TR-VPS
0.9776
0.6843
91.69
TR-only covariates: age, mechanism of injury, transfer status, ED systolic blood pressure, ED pulse, ED GCS motor score, max head AIS, max extremity AIS, congenital comorbidities VPS-only: PIM2 TR-VPS: TR model + VPS-PIM2 model
Mortality
ROC by Data Source 0.00
P = .0165
0.25
TR TR+VPS 0.50
1-Specificity VPS 0.75
1.00
Mortality
ROC by Data Source 0.00
P = .0165
0.25
TR TR+VPS 0.50
1-Specificity VPS 0.75
1.00
Mortality
VPS
Can we appropriately risk adjust without controlling for mechanism of injury? Injury severity?
Trauma Registry
Can we do better? Can we improve model fit? Improve accuracy? Efficiency?
Non-mortality outcomes
PCPC POPC PELOD Length of hospital stay Discharge to home (vs. rehab)
Hospital disposition
What are predictors of discharge home? What are predictors of discharge to a rehab facility?
What factors are most strongly associated with (poor) functional status?
III. Processes of Care
VTE prophylaxis Site A Site B Site C Site D Site E mechvte_hrs dvt_hrs
More…
Nutrition management Parenteral Enteral Daily bowel regimen C-spine clearance Alcohol and drug screening Alcohol counseling
Limitations
Non-mortality outcomes lack precision No quality of life measures Limited generalizability
Scope
75+VPS institutions w/ trauma ~40% ACS trauma centers ~60% state trauma centers 50+ centers can immediately merge data
Hurdles
Pediatric Trauma Assessment and Management Database
Conclusion
Combining databases is an innovative, feasible, cost-effective way to evaluate management practices and to explore critical questions related to pediatric trauma management.
Thank you
Special thanks to all trauma registrars and VPS coordinators at participating sites
Challenges…
…are worth it
Thank you
Questions?
VPS Predicted LOS PCPC POPC PIM2 PRISMIII PELOD Lab data Discharge status Pre-hospital data Initial vitals GCS Patient Outcomes Injury patterns Procedures Bedside procedures TR
Patient population
67 % male Mean age 7.2y (6.0) Race/Ethnicity 51% White 21% African American 7% Hispanic Payer 35% Private 48% Medicaid/Gov.
Injury characteristics
Mechanism of injury 32% Falls 25% MVC 4% Penetrating Intent 84% unintentional 14% assaults
TR
Place 31% residential Maximum Head AIS 15% AIS 4/5 43% AIS 3 Other max AIS 67% abd AIS 3-5 57% thoracic AIS 3-5 Injury Severity Score 13% ISS>25 22% ISS 16-25
TR
Pre-hospital & ED
Physiologic data 11% tachycardia* 3% hypotension* 9% GCS <9 EMS transport 42% ambulance 14% air
TR
*Age-based Physiologic data 29% tachycardia* 5% hypotension* 17% GCS <9 ED disposition 14% OR Transfer status
TR
ICU first hr & first 12 hrs
SBP 10% hypotension* Base excess -5.2 (4.2) Pupil reaction PF ratio
VPS
Physiologic/lab data BP, HR, RR, temp, pH P a O2, P a CO2 Hgb, WBC Plt, PT, PTT, bili K, Na, Ca, albumin, BUN, Cr Ventilation data Infection data
VPS
ICU course & outcomes
Baseline POPC 89% Normal 10% Mild/Mod 1% Severe Discharge POPC 34% Normal 57% Mild/Mod 4% Severe/Coma 5% Brain Death Intensivist (98%) 83% Concurrent care 5% Consulting only 10% Primary service PELOD
baseline, daily, POD
PRISM3 PIM2
VPS
Hospital disposition
ICU Length of stay Mean 2.8 (SD 5.0) Median 1.1 (.6-2.6) ICU disposition 69% floor, SDU 0.7% rehab 1.3% transferred
TR
Hosp length of stay Mean 7.3 (SD 10.9) Median 4 (IQR 2-8) Hosp disposition 82% home 11% rehab 2% transferred
VPS