Transcript ACCURACY OF PARAMEDIC DIAGNOSIS OF ACUTE CARDIOGENIC

ACCURACY OF PARAMEDIC DIAGNOSIS OF
ACUTE CARDIOGENIC PULMONARY OEDEMA
A prospective diagnostic audit of 1,334 patients
Emma Jenkinson*, Malcolm Woollard**,
Robert Newcombe†, Iain Robertson-Steel††
*Heartlands
Hospital/West Midlands Ambulance Service, **Faculty of Pre-hospital Care Research Unit,
†Medical Statistics Department, University of Cardiff, ††West Midlands Ambulance Service NHS Trust
United Kingdom
BACKGROUND
psychiatric
haematology
lungs
trauma
RESPIRATORY
DISTRESS
cardiac
renal
PATHOPHYSIOLOGY OF LVF
CPAP IN LVF
“…single greatest advance in the management of these
[LVF] patients in the past decade…”
Cohen Solal et al 1, 2004
CPAP is effective in patients with pulmonary oedema who
remain hypoxic despite maximal medical treatment
BTS Guidelines 2, 2002
Some UK ambulance services are looking to introduce
CPAP for paramedic use, one service already has.
PRE-HOSPITAL DIAGNOSIS OF ACPO
 Seven main studies 3
 Overall error rates 9-23%
 Additional study found 92% accuracy 4
 Paramedic identification of common lung sounds
found to be unreliable in 40% of cases 5
STUDY AIM
A prospective diagnostic audit to assess the
accuracy of paramedic diagnosis of acute
cardiogenic pulmonary oedema
DATA COLLECTION
 Prospective audit
 Population: patients brought to Heartlands (BHH)
by West Midlands Ambulance Service (WMAS)
 Publicised beforehand
 2 stages, 2 teams to allow for blinding
 An estimated 1,300 patients required
 Data collected between 4 Dec 05 until 31 Mar 06
DATA COLLECTION – STAGE 1
 WMAS PRFs searched to identify patients taken to BHH with:
 Diagnosis of ACPO OR
 Furosemide given OR
 Presenting complaint of respiratory distress OR
 Any of the following diagnoses:
 Acute asthma
 Croup
 SOB ?cause
 Exacerbation COPD
 Chest infection
 Pulmonary oedema
 Haemoptysis
 Excluded if seen by Dr
DATA COLLECTION – STAGE 2
 Demographics matched to hospital records to obtain:
 Emergency department (ED) diagnosis
 (Hospital discharge diagnosis)
 Investigator then unblinded
 Diagnoses matched
DATA ANALYSIS
 Two-by-two tables produced in SPSS:
 Positive or negative pre-hospital diagnosis of ACPO
 Positive or negative ED diagnosis of ACPO
 Results then entered into StatsDirect to calculate:
 Sensitivity
 Specificity
 PPV
 NPV
 PLR
 NLR
Proportion of patients with ACPO correctly
identified by ambulance staff as having ACPO
Proportion of patients without ACPO correctly
identified by ambulance staff as not having ACPO
By how much does the probability of having ACPO
increase with a positive pre-hospital diagnosis?
How much the probability of ACPO decreases with
a negative pre-hospital diagnosis of ACPO
RESULTS
Eligible patients (n=1,334)
No record (n=102)
GP referrals (n=34)
Transfer in (n=1)
Did not wait (n=16)
To primary care (n=19)
Patients seen by ED doctor (n=1,162)
ED diagnosis not recorded (n=7)
Complete data (n=1,155)
RESULTS
Complete data (n=1,155)
PRE-HOSPITAL DIAGNOSIS:
ACPO (n=59)
Not ACPO (n=1096)
ED DIAGNOSIS:
ED DIAGNOSIS:
ACPO
Not ACPO
ACPO
Not ACPO
(n=24)
(n=35)
(n=50)
(n=1046)
RESULTS
ALL PATIENTS
Pre-hospital diagnosis
ED diagnosis
ACPO
Not ACPO
Total:
ACPO
24
35
59
Not ACPO
50
1046
1096
Total:
74
1081
1155
95% Confidence intervals
 Prevalence
 Sensitivity
6.41%
32.43 %
5.06-7.98%
 Specificity
96.76%
95.53-97.73%
 Positive predictive value
40.68%
28.07-54.25%
 Negative predictive value
95.44%
94.03-96.60%
 Likelihood ratio of +ve result
10.02%
6.25-15.58%
 Likelihood ratio of –ve result
0.70%
0.58-0.80%
22.00-44.32%
RESULTS
PARAMEDIC
TECHNICIAN
6.39%
6.43%
4.70-8.46%
4.35-9.1%
46.67 %
10.34%
31.66-62.13%
2.19-27.35%
95.30%
99.05%
93.39-96.78%
97.59-99.74%
40.38%
42.86%
27.01-54.90%
9.90-81.59%
96.32%
94.14%
94.57-97.63%
91.54-96.14%
9.92%
10.91%
6.14-15.50%
2.78-40.98%
0.56%
0.91%
0.41-0.70%
0.74-0.97%
(n = 704)
 Prevalence
 Sensitivity
 Specificity
 +ve predictive value
 -ve predictive value
 LR +ve result
 LR –ve result
(n = 451)
RESULTS
ED diagnoses for patients with incorrect pre-hospital diagnosis of ACPO
No.
% total (95% CI)
RESPIRATORY
22
63% (45-79%)
Infection
8
23% (10-40%)
Chronic obstructive pulmonary disease
5
14% (5-30%)
Other respiratory disease
5
14% (5-30%)
Respiratory failure
3
9% (2-23%)
Asthma
1
3% (0-15%)
CARDIOVASCULAR
7
20% (8-13%)
Chest pain ?cause
3
9% (2-23%)
Syncope and collapse
1
3% (0-15%)
Angina/unstable angina
2
6% (0-19%)
Ischaemic heart disease – other
1
3% (0-15%)
OTHER
6
17% (6-35%)
TOTAL
35
100%
Diagnosis
CONCLUSIONS
 Sensitivity low, specificity high
 A positive diagnosis carried some predictive value
 If patients are treated for ACPO by pre-hospital
staff this is likely to be appropriate
 A large proportion of patients with ACPO are likely
to be missed
 Further training is required to improve diagnosis
REFERENCES
1.
Cohen Solal A. et al (2004) Traitement médical de l’insufficance cardiaque aigüe
décompensée. Annales de Cardiologie et d’Angéiologie 53: 200-208
2.
British Thoracic Society Standards of Care Committee (2002) Non-invasive ventilation
in acute respiratory failure. Thorax 57: 192-211
3.
Shapiro S.E. (2005) Evidence review: Emergency medical services treatment of
patients with congestive heart failure/acute pulmonary edema: do risks outweight the
benefits? J Emerg Nursing 31(1): 51-57
4.
Durham B., Aguilera P., Dale K., Neimen H. (1999) Accuracy of pre-hospital diagnosis
of primary respiratory distress. Acad Emerg Med 6(5): 474
5.
Widger H.N., Johnson D.R., Cohan S., Felde R., Colella R. (1996) Assessment of lung
auscultation by paramedics. Ann Emerg Med 28(3): 309-312