It's not about the alarms! -...
Transcript of It's not about the alarms! -...
keeping an eye on life.
Surveillance Monitoring System
It’s not about the alarms!
25-75% ofadverse eventsand preventabledeaths occuroutside of the ICU in unmonitored beds1
• Up to 60% of non-ICU beds are unmonitored2
• 84% of patients exhibit signs of deterioration within 8 hours preceding cardio-pulmonary arrest3
1 http://www.ihi.org/education/conferences/APACForum2012/Documents/I2_Presentation_Diagnostics_Haraden.pdf
2 AHA database, 2013 3 Schein RM et al. Clinical antecedents to in-hospital cardiopulmonary arrest.
Chest 1990;98:1388-92.
• Approximately 5% of hospital admissions have an unexpected transfer to the ICU*
• For every 1 hour increase in transfer delay, the odds of an in-‐hospital death increased 3%
• For pa?ents who survived un?l discharge, delayed transfer was associated with a longer length of stay.
Wendlandt, B et al. Association between ICU Transfer Delay and Hospital Mortality: A Multicenter Investigation (abstract). Journal or Hospital Medicine 2015:10 (suppl 2).
*Churpek,M et all. Multicenter Development and Validation Tool for Ward Patients. American Journal of Respiratory and Critical Care Medicine. Vol 190 No 6. Sept 15, 2014
Rapid Response Systems
ICU
General Floor
Professor Kenneth Hillman, AO
Conditional versus Surveillance Monitoring
Condi6onal Monitoring (ICU)
• Pa?ent has risk factors
• Monitoring as ordered
• Specialized monitoring
• Targeted measures (cardiac telemetry for cardiac pa?ents)
• High-‐risk popula?on
• Special wards
Surveillance Monitoring (General Floor)
• Environment has risk factors
• Monitoring as standard of care
• General monitoring
• Mul?-‐parameter measurements (HR, RR, BP, SpO2, etc)
• Lower-‐risk popula?on
• General care wards
Optimizing Alarms for Surveillance Monitoring
Objectives: 1. Detect patient deterioration 2. Alarm only on clinically actionable events
• Requires beside intervention • Reverses an emerging harmful event
3. Build large data base • Optimize alarm configuration settings • Develop strategies to reduce false alarms • Develop new “smart” alarm algorithms
Optimizing Alarms for Surveillance Monitoring Methods: 1. Capture high fidelity physiologic data in
intended application across multiple hospitals • Capture and store data at the hospital • De-identify PHI • Transfer and store data in cloud
repository 2. Characterize distribution of each vital sign
under actual clinical use 3. Perform “what if” analysis across multiple
alarm configurations • Optimize threshold settings • Optimize alarm delays
275,000 hours of pa?ent data
Predictive Analytics Alarm Configura6on
Alarm Performance De-‐iden?fy PHI
Distribution of SpO2 values (general care wards)
90% 85%
Median = 96%
250 patients 8,400 hrs
Combined impact of threshold and delay settings on SpO2 alarm
• Delays have the greatest overall benefit in reducing alarms
• 85% SpO2 alarm threshold represents 4 sigma from SpO2 median
• 60 sec delay lowers alarm rate to 2.8 alarm / pt / day
Data based on one institution, 15k hrs
National Benchmark
• 275k+ hours
• 9500 pa?ent sessions
• 16 Hospitals
May 2015 Aggregate Analysis
Parameter Alarms
SpO2 3.01 Cardiac Rate 2.96
Respira6on Rate 0.55 NIBP 0.04 cNIBP 0.86 Total 7.42
Parameter High Alarm Low Alarm
Delay (sec)
SpO2 -‐-‐ 85 60 Cardiac Rate 150 30 30
Respira6on Rate 35 4 120 Systolic Blood Pressure 190 -‐-‐ 120 Mean Arterial Pressure -‐-‐ 60 60 Diastolic Blood Pressure -‐-‐ -‐-‐ -‐-‐
Jan – May 2015 Partner Hospital
Parameter Alarms
SpO2 1.84 Cardiac Rate 1.38
Respira6on Rate 0.47 NIBP 0.04 cNIBP 1.31 Total 5.03
• 26 bed post surgical care unit • Policy to surveillance monitor all pa?ents • Data capture Feb 9, 2015 – May 31, 2015
delays thresholds Alms/pt/day
Best Practice Hospital
SpO2 OFF/90 OFF/85 1.14PR 60/60 150/29 0.3HR 15/15 150/29 2.69CR 60/60 150/29 1.27RR 120/120 35/4 0.38BP 240,OFF,OFF/OFF,OFF,60 200,OFF,OFF/OFF,OFF,58 0.06
cNIBP 240,OFF,OFF/OFF,OFF,60 200,OFF,OFF/OFF,OFF,58 0.43Total 3.29
SpO2 OFF/90 OFF/85 1.14PR 60/60 150/29 0.3HR 15/15 150/29 2.69CR 60/60 150/29 1.27RR 120/120 35/4 0.38BP 240,OFF,OFF/OFF,OFF,60 200,OFF,OFF/OFF,OFF,58 0.06
cNIBP 240,OFF,OFF/OFF,OFF,60 200,OFF,OFF/OFF,OFF,58 0.43Total 3.29
Leading Indicator Report
• Generated for every ac?ve account every 2 weeks
• Reports alarm rates and # of sessions
• Nega?ve trends reported to the hospitals
• Correc?ve ac?on implemented before nuisance alarms are reported
Thank you [email protected]