Abstract
Introduction
The aim of the present paper is to investigate differences in the
apnea/hypopnea index (AHI) during two consecutive nights in the sleep laboratory. The
results of this study are based on data recorded in the SIESTA project. A new
apnea/hypopnea detection software, which was developed recently for the Somnolyzer
24x7, will be introduced.
Method
The detection algorithm is based on 4 polysomnographic signals: oxygen
saturation (SaO2), nasal airflow, movement of the chest wall and of the abdomen .
Oxygen saturation is first resampled to 4 Hz. Subsequently, peaks of this signal are
determined in order to extract intervals of oxygen de- and re-saturation. A similar
technique is used to detect changes in nasal airflow. In a first step, baseline drifts are
subtracted from the signal by highpass filtering with a 0.05 Hz edge frequency. Intervals
of decreased airflow are then calculated by comparing the signal with a smoothed version
of itself. For further processing, intervals with reduced airflow are assigned to two
classes: an amplitude decrease of more than 50% and more than 80%. The chest and the
abdominal movement channels are treated similarly. The intervals detected in the airflow
channel are projected onto these channels and are further classified into the two
categories described above. The detection of apnea events is accomplished by running
through a decision tree structure. Based on the decrease in airflow, as well as on the chest
and abdominal movement signals, the specified intervals are classified as either central,
obstructive, or mixed apneas or hypopneas. In the present analysis 51 subjects (44 males
and 7 females) recorded in 2 consecutive nights were included. Subjects' mean age was
51+/-10 years. Differences between AHIs obtained in the first and in the second night
were statistically compared by means of a paired t-test.
Results
The AHI was 35.91 in the first and 36.09 in the second night, with a higher
variance in the first night. The paired t-test did not reveal a significant difference between
the two nights (t t=-0.045; p=0.964). In 5 recordings a critical AHI value of 12 was not
reached in the first, but in the second night. Four patients did not show an AHI of more
than 12 in either night.
Conclusion
Our results confirm previous reports that with a high AHI measured already
during the first night, sleep apnea might be diagnosed without a second night of
polysomnography, but that in cases of an only slightly increased AHI, a second night is
absolutely necessary. The recently developed automatic apnea detection method has
proven to be a robust tool for the evaluation of sleep apnea events.
Go back