Evaluation of the add-on NOWAPI ® medical device for remote monitoring of compliance to Continuous Positive Airway Pressure and treatment efficacy in obstructive sleep apnea

Patients were mainly men (18/22), aged 61 ± 13 years and with BMI of 29.6 ± 4.5 kg/m
²
(mean ± SD). Their baseline characteristics are outlined in Table 1. Reasons for the in-hospital night respiratory polygraphic recording were: unrefreshing
sleep (10/22), recurrent awakenings (8/22), disturbed sleep (5/22), excessive daytime
sleepiness (5/22), poor compliance to CPAP treatment (2/22) and residual AHI 20/h
(1/22). From available prior home CPAP reports (n = 21), residual AHI was 5.8 ± 5.5/hr
and in only three patients was ?10/hr, whereas 10 such patients were requested per
protocol. Two patients with major protocol deviations (less than 4 h of respiratory
polygraphic recordings, i.e. 0 and 2.25 h respectively) were not included in the main
analysis. Interfaces included 12 face masks, 8 nasal masks and 2 nasal pillows. 20
patients had an auto-adjusting CPAP device and two had a conventional CPAP device.
6 different CPAP devices were used [S8 AutoSet Spiritâ„¢ II (ResMed), S9 AutoSetâ„¢ (ResMed),
REMstar Auto A-Flexâ„¢ PR One (Philips Respironics), REMstar Auto M Series with A-Flex
(Philips Respironics), GoodKnight
^®
420 (Tyco healthcare), KXS-Bump (Kaerys)]. Fourteen patients had a humidifier and
eight none. Four patients rose from bed and three patients removed their interface
during the recording night.

Table 1. Characteristics of the study population (full analysis set, n = 22)

Considering the overall recording night, the median CPAP treatment duration estimated
by NOWAPI
^®
and from polygraphy was of 471 min (ranging from 11 to 570) and 479 min (ranging from
405 to 585), respectively (PPS). The median [Q1; Q3] absolute difference in CPAP treatment
duration between NOWAPI
^®
and polygraphy was of 1.0 min [0.0; 12.0], corresponding to a relative absolute difference
of 0.21 % [0.0; 2.2] (PPS). The maximum absolute difference was of 439 min (97.6 %
in relative absolute difference). Figure 3 presents for each patient the CPAP treatment duration measured by NOWAPI
^®
against the one measured by polygraphy. It shows that NOWAPI
^®
measured the exact same treatment duration as polygraphy in all but four patients.

Fig. 3. Comparison of overnight CPAP treatment duration (minutes)—Respiratory Polygraph vs
NOWAPI
^®
(PP data set, n = 20). Crosses represent the 16 patients with a perfect agreement in CPAP treatment duration measurement
between NOWAPI
^®
and polygraphy, all crosses being aligned on the 1:1 fit line. The circles represent
the four outliers. Detailed NOWAPI
^®
data files of patients #19 and 20 are presented in Fig. 4

Examination of NOWAPI
^®
detailed data files was conducted to further analyze the CPAP treatment duration results
and in particular the outliers. Detailed data files were available in only 18 of the
20 patients of the PPS. An example of NOWAPI
^®
detailed data is shown in Fig. 4: flow rate determination (i.e. treatment detection) is allowed by calculating the
difference between the two pressure signals (pink and blue curves). If this difference
(yellow curve) varies, it means that there is a flow rate, like what is seen in patient
#20. In patient #19, the two pressure signals are always superimposed: this finding
is compatible only with two situations, either no flow rate in the patient circuit,
either reverse connection of the NOWAPI
^®
on the patient circuit. Review of the 18 NOWAPI
^®
detailed data files showed no variation in differential pressure (indicating with
high probability the reverse connection of the NOWAPI
^®
on the patient circuit, gas flow in the wrong sense through the device, during the
recording night) in the 4 outlier patients. NOWAPI
^®
detailed data files of all other 14 patients showed the expected variation in differential
pressure.

Fig. 4. Example of NOWAPI
^®
detailed data in two typical patients. The device measures two pressure signals (upstream
restriction and at the site of restriction, pink and blue curves, left y–axis). Variations in the difference between two pressures (differential pressure, yellow curve, right y–axis) indicate a flow rate is detected. Panel a shows detailed data in patient #20, with variation in differential pressure indicating
detection of a flow rate in the patient circuit, i.e. a CPAP treatment. Panel b shows detailed data in patient #19. The two pressure signals are always superimposed,
there is no variation in differential pressure. This finding is compatible only with
two situations, either no flow rate in the patient circuit, either reverse connection

A post hoc analysis was therefore conducted on the 14 patients having a detailed data
file indicating correct connection: Considering the overall recording night, the median
CPAP treatment duration estimated by NOWAPI
^®
and from respiratory polygraphy was of 478 min (ranging from 405 to 570) and 479 min
(ranging from 405 to 585), respectively. Median [Q1; Q3] absolute difference in CPAP
treatment duration between NOWAPI
^®
and respiratory polygraphy was of 0.0 min [0.0; 1.0], corresponding to a relative
difference of 0.0 % [0.0; 0.24] overall on the recording night for this subgroup.
The maximum absolute difference was of 15 min (2.6 % in relative difference).

Evaluation of the performance of the NOWAPI
^®
in apnea and hypopnea detection was therefore conducted only in patients for whom
NOWAPI
^®
detailed data file examination indicated a correct device connection. Differences
between numbers of apneas and hypopneas estimated by NOWAPI
^®
and from polygraphic data interpretation are presented for each of the common 15-min
recordings in Fig. 5 (panels A, B and C). Agreement between NOWAPI
^®
and polygraphy was within two apneas or less in most records (427/452, i.e. 94 %),
and similarly within two hypopneas or less in most records (408/452, i.e. 90 %). Underestimation
of more than two respiratory events par NOWAPI
^®
(20/452 and 37/452 of records in apnea and hypopnea detection respectively) was most
frequent than overestimation of more than two respiratory events (5/452 and 7/452
of records in apnea and hypopnea detection respectively). Considering the overall
recording night, the median [Q1; Q3] absolute difference in AHI was 2.2 [1.1; 3.4]
events per hour between NOWAPI
^®
and polygraphy. Figure 5 (Panel D) shows residual AHIs measured by NOWAPI
^®
plotted against respiratory polygraphic data interpretation for the 14 patients overall
on the recording night, and confirmed NOWAPI
^®
underestimation of residual respiratory events, in an order of magnitude of about
two events per hour.

Fig. 5. Differences between numbers of apneas and hypopnoeas (Panel a), apneas (Panel b) and hypopneas (Panel c) estimated by NOWAPI
^®
and by respiratory polygraphy are presented for each of the common 15-min periods
(n = 452) of the recording night. The x-axis plots the difference [NOWAPI
^®
minus Respiratory Polygraph] and the y–axis the number of records. Panel d presents the Apnea-Hypopnea Index (events per hour) by patient for the overall recording
night, estimated by NOWAPI
^®
plotted against polygraphy with a 1:1 fit line. PP data set with detailed NOWAPI
^®
data files indicating correct connection (n = 14)

From the sleep agenda, TST was 433.0 ± 105.8 min (around 7.2 h), SOL was 32.8 ± 27.8 min,
WASO was 34.4 ± 49.5 min and TIB was 501.9 ± 51.2 min (around 8.3 h) (n = 22) (mean ± SD).

Figure 6 shows patient opinion on NOWAPI
^®
. Patients were globally highly satisfied of the design of the device (shape, size,
color). Diode indicator, included to provide feed-back to the patient on his compliance,
obtained the lowest satisfaction rating. No patient reported that the traffic light
indicator was very helpful (100 % rating) in indicating the quality of the treatment
and 42 % of patients reported that it was not helpful at all (0 % rating). Conversely,
the patients quite high-rated disturbance during the night, 8 % of patients having
reported high disturbance (100 % rating) and 17 % of patients reported no disturbance
(0 % rating).

Fig. 6. Patients’ opinion on NOWAPI
^®
(n = 12). Results are mean ± SD. The self-administered satisfaction questionnaire
included three questions on the general aspect of the device and two questions on
the light indicator. Answers were given using a satisfaction scale (0, 20, 40, 60,
80, 100 %), 0 % being the poorest satisfaction and 100 % the maximum

No adverse event was reported during the study.