Description

In developing countries it is important for local caregivers to be able to identify infants and small children with hypoxemia associated with acute lower respiratory tract infections based solely on clinical findings. Hypoxemic infants have a much higher mortality rate than nonhypoxemic infants and require more aggressive therapy. The study was done in Gambia in conjunction with the WHO and London School of Hygiene and Tropical Medicine.


Definition for hypoxemia: arterial oxygen saturation < 90% using a pulse oximeter (corresponds to an arterial oxygen tension of 60-70 mm Hg, depending on age, altitude, blood pH, temperature and other factors).

 

Endpoints: To identify those infants to receive oxygen therapy and referral to the local hospital.

 

Patient population in study of Usen et al: based on enrollment in a Hemophilus influenzae vaccination study:

(1) Exclusions: age < 2 months (not candidates for vaccine), Down's syndrome, structural heart disease

(2) 1,072 children, with 5.9% having hypoxemia

(3) mean age: 8 months (non-hypoxemic) and 9 months (hypoxemic)

 

 

Hypoxemia if 1 or more of the following is present

Finding

Dyke

Onyan-go

Usen (1)

Usen (2)

Weber

WHO

inability to feed or drink

 

 

 

 

 

x

cyanosis

x

 

x

 

x

x

respiratory rate in breaths/minute

>= 90

>= 70

 

>= 90

 

>= 70

severe chest indrawing

 

inter-costal

 

 

 

x

grunting respirations

x

x

 

 

 

 

abnormally sleepy

x

 

 

 

 

 

head nodding

 

 

x

x

 

 

inability to cry

 

 

 

x

x

 

from Table 2, Usen page 89

 

Criteria

Sensitivity

Specificity

Positive Predictive Value

Negative Predictive Value

Dyke

73%

72%

13.9%

97.7%

Onyango

84.2%

63.3%

11.3%

98.6%

Usen (1)

69.3%

80.4%

17.8%

97.6%

Usen (2)

69.8%

78.6%

16.9%

97.7%

Weber

71.4%

77.1%

16.3%

97.7%

WHO

80.9%

62.5%

11.9%

98.1%

from Table 2, Usen page 89

 

The criteria appear fairly good for excluding hypoxemia (high negative predictive value).

 

Independent predictors for hypoxemia using logistic regression model (Usen et al):

(1) cyanosis

(2) rapid respiratory rate

(3) grunting respirations

(4) head nodding

(5) absence of a history of fever

(6) no spontaneous movements during the examination

 


To read more or access our algorithms and calculators, please log in or register.