Evidence Base: Pulmonary

Accuracy in Diagnosis of Pneumonia

Question 1: Does this patient have community-acquired pneumonia?

History

History/Symptoms	LR+	LR-
Dyspnea	1.4 NS in 2 studies	0.67 NS in 2 studies
Cough	1.8 NS in 2 studies	0.31 NS in 2 studies
Sputum Production	1.3	0.55 NS in 2 studies
Fever	1.7 - 2.1	0.59 - 0.71
Chills	1.3 - 1.7	0.70 - 0.85
Night Sweats	1.7	0.83
Myalgias	1.3 NS in 1 study	0.58 NS in 1 study
Sore Throat	0.78 NS in 1 study	1.6 NS in 1 study
Rhinorrhea	0.78 NS in 1 study	2.4 NS in 1 study
PMH: Asthma	0.10	0.85
PMH: Immunosuppression	2.2	0.85
PMH: Dementia	3.4	0.94
NS = result not significant

Adapted from Metlay JP, et al.

back to top

Question 2: Does this patient have pneumonia

Examination

Agreement between examiners (A, B, C) on findings reflected by Kappa values
		A vs B	B vs C	A vs C
Rales	R Lung	.24	.49	.65
	L lung	.35	.64	.51
Rales lat decub	R Lung	.32	.39	.39
	L lung	.23	.47	.23
Auscultatory Percussion	R Lung	-.04	.28	.10
	L lung	0	0	.45
Pneumonia Dx	Kappa	.18	.31	.43
	% Agreement	60	69	72
Agreement between examiners for diagnosis of pneumonia is presented as kappa value and % agreement

Permission to reproduce granted by author. Wipf, et al.

Three examiners (A,B,C): sensitivity and specificity of Physical Findings for each examiner
		A		B		C
		Sens	Spec	Sens	Spec	Sens	Spec
Rales	R Lung	.54	.61	.75	.39	.56	.53
	L lung	.67	.70	.50	.63	1.0	.79
Rales lat decub	R Lung	.69	.70	.56	.61	.44	.47
	L lung	.33	.88	.25	.93	1.0	.96
Auscultatory Percussion	R Lung	.39	.91	.06	.97	.44	.88
	L lung	.33	.90	0	1.0	.50	.96
Pneumonia Dx	Overall	.69	.65	.47	.58	.50	.75

Permission to reproduce granted by author. Wipf, et al.

Three examiners (A,B,C): Likelihood Ratios for Physical Findings for each examiner
		A		B		C
		LR+ (95% CI)	LR- (95% CI)	LR+ (95% CI)	LR- (95% CI)	LR+ (95% CI)	LR- (95% CI)
Rates	R Lung	1.4 (.72-2.68)	.75 (.39-1.44)	1.2 (.83-1.82)	.64 (.25-1.82)	1.2 (.56-2.57)	.83 (.35-1.97)
	L Lung	2.2	.47	1.4	.79	4.8	0
Rales lat decub	R Lung	2.3 (1.22-4.34)	.44 (.19-1.03)	1.4 (.78-2.64)	.72 (.39-1.34)	.08 (.35-1.97)	1.2 (.55-2.57)
	L Lung	2.8 (.67-11.33)	.76 (.42-1.35)	3.6 (.69-18.39)	.81 (.54-1.21)	25 (3.53-177.5)	0
Auscultatory Percucssion	R Lung	4..3 (1.2-15.59)	.67 (.43-1.05)	2 (.13-31.08)	.97 (.85-1.11)	4 (.83-1.17)	.64
	L Lung	3.3 (.76-14.37)	.74 (.42-1.32)	-	1 (1-1)	12.5 (1.13-137.9)	.52 (.13-2.09)
Pneumonia Dx	Overall	2 (1-3.9)	.48 (.21-1.06)	1.1 (.55-2.28)	.91 (.52-1.62)	2 (.6-6.64)	.67 (.31-1.43)
Permission to reproduce granted by author. Wipf, et al.

back to top

Compiled Data from Several Studies

Examination	LR+	LR-
Vital Signs
Resp Rate > 20 breaths/min	1.2	0.66
Resp Rate > 25 breaths/min	1.5-3.4	0.78-0.82
Resp Rate > 30 breaths/min	2.6	0.80
Heart Rate > 100 breaths/min	1.6-2.3	0.49-0.73
Heart Rate > 120 breaths/min	1.2	0.66
Temperature > 37.8 C (100 F)	1.2	0.66
Pulmonary Examination
Dullness to percussion	2.2-4.3	0.79-0.93
Decreased breath sounds	2.3-2.5	0.64-0.78
Crackles	1.6-2.7	0.62-0.87
Bronchial breath sounds	3.5	0.90
Rhonchi	1.4-1.5	0.76-0.85
Egophony	2.0-8.6	0.76-0.96
Adapted from Metlay JP, et al.

back to top

Prediction Rules

Since no one symptom nor examination finding is highly accurate, authors have developed prediction rules that combine several findings.

A simple decision rule suggested by Gennis, et al.
Obtain chest x-ray for patients suspected of having pneumonia with at least 1 vital sign abnormality.
Another decision rule by Heckerling, et al.
Determine the number of findings present: Absence of asthma Temperature > 37.8 C (100 F) Heart rate > 100 beats/min Decreased breath sounds Crackles Determine pretest probability of pneumonia in the population. In most outpatient settings, ~5% of patients presenting to primary physician with cough are diagnosed with pneumonia. In this setting: # of findings Predicted probability of pneumonia 2 3% 3 10% 4 25% 5 50% Heckerling PS, et al.

back to top

Conclusions

1. Physician agreement about the presence or absence of pneumonia in patients with respiratory illness is relatively low.
2. No single symptom, historical feature nor physical examination finding is highly accurate in predicting pneumonia.
3. Prediction rules using combinations of findings are helpful in ruling out pneumonia but chest x-ray is required to accurately diagnose pneumonia.
4. A screening pulmonary physical examination may include percussion and auscultation for crackles. If abnormalities are detected or if pneumonia is strongly suspected, one may consider maneuvers such as egophony.

back to top