Previous studies have reported great difficulty in recording lung soun
ds from neonates and have found conflicting results. We studied lung s
ounds in neonates during the inspiratory phase of the respiratory cycl
e as monitored by inductive plethysmography (A) and by a pneumotachogr
aph and a face mask (B) which added a dead space of 12 mt. Sixteen ter
m babies were tested 12 hr to 6 days (median 45 hours) after birth. Lu
ng sounds were recorded and then analysed using overlapping and non-ov
erlapping fast Fourier transforms. The two methods of analysis showed
a difference in intensity but not in frequency. Fourteen babies provid
ed enough breaths for comparison; a total of 596 inspirations were ana
lysed. The intensity of lung sounds on occasion B was higher in all bu
t two babies with a mean BIA ratio of 2.4. The mean (SD) power on occa
sions A and B was 13.9 (8.5) mW and 26.9 (21.0) mW, P = 0.02, respecti
vely. In all but 4 babies the B/A ratios of the median (f(50)) and 90t
h centile (f(90)) frequencies were scattered randomly within 20% of un
ity. The mean (SD) f(50) on occasions A and B was 205.5 (51.1) Hz and
225.8 (32.3) Hz, P = 0.10, respectively; the mean f(90) was 370.3 (91.
0) Hz and 396.1 (67.8) Hz, P = 0.25, respectively. Linear regression s
howed that there is a third-order polynomial relationship between soun
d intensity and air flow at the mouth. A weaker positive association e
xists between frequency and air flow, showing that the median and 90th
centile frequencies approach an asymptote as flow increases. We concl
ude that adding a dead space increases the intensity of lung sounds in
neonates due to increased flow, but has no consistent effect on frequ
ency content. Pediatr Pulmonol, 1995; 19:348-354. (C) 1995 Wiley-Liss,
Inc.