Prospective Study to Revise the Ottawa Heart Failure Risk Scale

Prospective Study to Revise the Ottawa Heart Failure Risk Scale (OHFRS)
CAEP
Edmonton 2015
Ian Stiell MD Shawn Aaron MD
Robert Brison MD
Alan Forster MD
Jeffrey Perry MD George Wells PhD
Catherine Clement RN Bjug Borgundvaag MD
Lisa Calder MD Andrew McRae MD
Brian Rowe MD
Department of Emergency Medicine
Ottawa Hospital Research Institute
University of Ottawa, Ottawa, ON
Queens University, Kingston, ON
University of Alberta , Edmonton, AB
University of Toronto, Toronto, ON
University of Calgary, Calgary, AB
Funded by Canadian Institutes of Health Research
No Conflicts to Declare
Acute Heart Failure in the ED:
The Clinical Problem
Very common presentation
Potential for poor outcomes
Many HF patients do not need hospital admission
Little evidence to guide disposition decisions
Need risk stratification tool to aid rational and safe admission decisions
Derivation of Ottawa Heart Failure Risk Scale:
Acad Emerg Med 2013
Goal: Improve and standardize admission practices for HF
Methods:
Prospective cohort study at 6 EDs
N=559 with 38% admitted
Risk scale derived by logistic regression
Original Ottawa Heart Failure Risk Scale
Objectives:
Prospective Validation Study (2012‐2014)
1. To prospectively and clinically evaluate OHFRS in a new patient population  2014
2. To improve upon OHFRS by re‐evaluating predictor variables and developing a more concise model  2015
Methods ‐ 1
Design: Prospective cohort study
Setting: 6 EDs of large, tertiary care Canadian hospitals
Subjects: Acute HF ‐ admitted and discharged
Excluded:
O2 Sat < 85% on RA, HR > 120, SBP < 85
chest pain or STEMI
Ethics: approved +/‐ informed consent
Standardized Assessment: variables from history, physical
routine labs, troponin, CXR, ECG
quantitative NT‐ProBNP
3‐minute walk test
Methods ‐ 2
Primary Outcome Measure ‐ serious adverse event (SAE): Death <30 days
Intubation or NIV after admission
MI
Admission to monitored unit, or Relapse back to ED with admission within 14 days
Data Analysis: Univariate association with SAE
Harmonized troponin by URL
Logistic regression to revise model
Rounded β‐coefficients to create points
Patient Flow
Patients Seen N=4,999
Patients Eligible N=1,869
Patients Enrolled N=1,100
Final Study Cohort N=1,100
Patients Excluded N=3,130
‐ No clear increase SOB N=788
‐ CXR shows no HF N=394
‐ Nursing home N=376
‐ Outside study hours N=387
‐ Confusion, dementia N=276
‐ Enrolled< 2 months N=200
‐ Other N=709
Patients Not Enrolled N=769
Patient Characteristics
Characteristic (%)
Age in years, mean
Range Male (%)
Ambulance arrival (%)
Duration of symptoms, hours, mean
CTAS level, median
Past medical history (%)
Heart failure
MI/Angina
Intubation for resp distress
N=1,100
77.7 50‐104
53.1
44.4
65.5
2
76.7
39.2
1.1
Patients Enrolled by Hospital (N=1,100)
Foothills MC 12.8%
Kingston General 12.6%
Ottawa Hospital, General Campus 14.6%
U Alberta 14.0%
Ottawa Hospital, Civic Campus 30.5% Mount Sinai 15.6%
Patient Characteristics (N=1,100)
Current Cardiac Meds (%)
Diuretics
Beta blockers
Statins
Ace Inhibitors
Calcium Channel Blockers
Anti‐arrhythmics
Amiodarone
Sotalol
Propafenone
75.0
68.9
62.4
42.5
36.3
6.8
6.0
0.4
0.2
Patient Outcomes (%)
N=1,100
Admitted to Hospital (N=629)
Critical Care Unit
Non‐Invasive Ventilation
Intubation
MI after admission
Death after admission
Death after discharge
57.2
9.7 3.8
0.8
3.5
4.0
1.3
Discharged from ED (N=471)
Relapse back to ED Relapse and admitted
Death within 30 Days 42.8
24.2
10.6
1.7
Serious Adverse Events 25
19.4%
20
SAEs (%)
15.6%
15
10.2%
10
5
0
Total (N=1,100)
Admitted (N=629)
Discharged (N=471)
Univariate Correlation with SAEs for 10 OHFRS Criteria
Criterion (%)
History of stroke or TIA
History of intubation
HR on ED arrival >110 Room air SaO2 <90
ECG acute ischemia
Urea > 12 mmol/L
Serum CO2 >35 mmol/L
Troponin >0.10 ug/L
NT‐ProBNP >5,000 ng/L
Walk test failed / too ill SAE No SAE P‐Value
(N=170) (N=930)
15.9
15.4
0.87
1.2
1.0
0.80
20.1
11.1
<0.05
32.9
22.2
<0.05
9.5
0.3 <0.0001
41.8
27.2 <0.001
5.4
1.4 <0.001
33.3
14.3 <0.0001
63.3
49.0 <0.05
70.3
53.1 <0.0001
Univariate Correlation with SAEs for Other Variables
SAE No SAE P‐Value
(N=170) (N=930)
On Anti‐arrhythmic
10.6
5.6
<0.05
Ejection Fraction, mean% N=152,739
43.4
46.3
<0.05
<30% 21.7
19.5
0.53
ED Treatment
IV NTG
9.4
1.8 <0.0001
BiPAP
17.1
5.0 <0.0001
Initial or repeat Troponin
>URL
71.0
56.8 <0.001
> 3x URL
39.1
18.7 <0.0001
30.2
9.9 <0.0001
> 5x URL
NT‐ProBNP
37.8
25.5
<0.05
>10,000 ng/L
27.7
13.1 <0.0001
>15,000 ng/L
Variable (%)
Independent Predictors of SAE from Logistic Regression P‐value to enter <0.25, remove >0.10
Hosmer‐Lemeshow Goodness‐of‐fit p‐value = 0.90
Area under ROC curve = 0.72
Revised Ottawa Heart Failure Risk Scale
Classification Performance & Potential Admissions for OHFRS Cutpoints
** model developed for 1,045 patients without missing values
Observed vs Expected Probability of SAE Score Internal validation ‐ very accurate x 1,000 replications using bootstrap
Discussion
Much higher admission rate
Some component variables did not perform well
NT‐proBNP not useful
Used troponin as multiples of URL
Revised OHFRS more concise model Good internal validation
Limitations
Requires further clinical validation
Physicians often neglected re‐assessment/walk test
Conclusions: Revised OHFRS
Consists of 6 simple, bedside variables Estimates short‐term risk of SAE in acute heart failure Will assist MDs in making rational ED disposition plans
Should improve and standardize admission practices, diminishing both unnecessary admissions for low‐risk patients, and unsafe discharge decisions for high‐risk patients
Ultimately lead to improved safety for patients and more efficient use of hospital resources
Details of Troponin Assay by Hospital Site Methods
3‐Minute Walk Test:
RN or respiratory therapist
Exclude if SaO2 <90% or HR>120
Measure pulse oximetry
Primary Outcome Measure ‐ serious adverse event (SAE): Death <30 days
Intubation or NIV after admission
MI
Admission to monitored unit, or Relapse back to ED with admission within 14 days
Missed Patient Characteristics (N=769)
Age in years, mean
Range Male (%)
Admitted
Hospital (%)
Ottawa H Civic, Ottawa, ON
Ottawa H General, Ottawa, ON
Foothills MC, Calgary, AB
Mount Sinai H, Toronto, ON
University Alberta H, Edmonton, AB
Kingston General H, Kingston, ON
77.7
50‐102
51.5
60.2
33.3
25.5
18.7
9.3
6.9
6.4
Classification Performance of OHFRS
without NT‐proBNP (N=1,100)
SAE
OHFRS
>2
<2
Sensitivity
Specificity
Admission
Yes
No
121
49 410
520
71.2%
55.9%
48.3%
Classification Performance of OHFRS
with NT‐proBNP (N=662)
SAE
OHFRS
>2
<2
Sensitivity
Specificity
Admission
Yes
No
95
24 336
229
79.8%
40.5%
63.0%
Physician Comfort with Use of the OHFRS
Uncomfortable 8.5%
Very Uncomfortable 3.5%
Very Comfortable 17.0%
Neutral 21.8%
Comfortable 41.5%
**84 no response
Patient Flow
Patients Seen N=4,999
Patients Eligible N=1,869
Patients Enrolled N=1,100
Final Study Cohort N=1,100
Patients Excluded N=3,130
‐ No clear increase SOB N=788
‐ CXR shows no HF N=394
‐ Nursing home N=376
‐ Outside study hours N=387
‐ Confusion, dementia N=276
‐ Enrolled< 2 months N=200
‐ Other N=709
Patients Not Enrolled N=769
Patients Lost to Follow‐up
N=2