Simple Automation Reduces False Positive Rate in Perfusion Index Critical Congenital Heart Disease (CCHD) Screening

This submission has open access
Submission ID :
ESPR264
Submission Type
Submission Topic
Abstract: :

Background


Pulse oximetry screening fails to diagnose an estimated 900 US newborns with critical congenital heart disease (CCHD) annually. The most commonly missed defect is coarctation of the aorta (CoA). These defects cause poor perfusion rather than hypoxemia. Perfusion index (PIx), a measurement of systemic perfusion, may improve detection of defects such as CoA. However, it has been noted that PIx has inherent variability prompting concern that it may be unreliable or difficult for bedside value assignment by clinicians. 


Objective

Assess the impact of a low fidelity automated screening algorithm that combines pulse oximetry measurements of SpO2 and PIx for CCHD screening.


Designs/Methods

We prospectively enrolled newborns (< 8 d) with and without congenital heart disease (CHD) and recorded simultaneous pre and postductal pulse oximetry data over 5 minutes. We analyzed these recordings with an automated screening algorithm that only utilized motion-free PIx and SpO2 values. 


The automation algorithm calculated the PIx variation and omitted data with high variation and detected the most recurrent PIx values and averaged SpO2 values over the motion free packets. Single manual SpO2 and PIx values were assigned at "bedside" using a stable waveform without artifact for 10 secs. Cut offs for postductal or post+preductal PIx < 0.4­ and 0.5 (5th and 10th percentiles using normal values) and SpO2 Kemper algorithm were utilized to determine pass/fail rates and compared between the manual and automated algorithms using McNemar's test. Babies were confirmed to be healthy after at least 3 months old.


Results

142 newborns (88 without CHD, 33 with CCHD and 21 with non-critical CHD,) were included in this analysis. When applying the standard bedside SpO2 alone algorithm all healthy babies passed the SpO2 screen. Of the CCHD, 13 (39%) passed SpO2 alone screen (6 of the 13 or 46% had systemic obstruction). Using PIx 0.4, a newborn with critical CoA that would have been missed by SpO2 alone would have been detected. The echocardiogram even underestimated the severity of the obstruction and the infant was initially discharged before having surgery but returned requiring prostaglandin and surgery. By increasing PIx threshold to < 0.5 and with including preductal PIx, additional infants with CCHD and non-critical CHD undetected by SpO2 screen could be detected; however, the false positive rate also increased (Table 1). Automation significantly reduced the false positive at PIx 0.4 threshold but did not significantly decrease it at 0.5 PIx threshold. Automation of just the SpO2 component also would have diagnosed 3 newborns with CHD undetected by bedside SpO2 (1 critical, 2 non-critical).


Conclusion(s)

Automation that utilizes multiple motion free PIx values may lower false positive rate if PIx is introduced into CCHD screening. However, the PIx threshold is yet to be identified as lowering the threshold reduces the number of false positives but also misses CCHD. 


Optional insertion of tables and or figures :
If the file does not load, click here to open/download the file.
University of California, Davis
University of California, Davis
University of California, Davis
University of California, Los Angeles
Northwell Health - Cohen Children's Medical Center
Northwell Health - Cohen Children's Medical Center
University of California, San Francisco
University of California, San Francisco
Sutter Health
University of California, Davis

Similar Abstracts by Type

Submission ID
Submission Title
Submission Topic
Submission Type
Corresponding Author
ESPR157
Clinical Research
Original science
Aditya Chhikara
ESPR302
Epidemiology
Original science
Natasha Jouk
ESPR74
Clinical Research
Original science
Alexandra Mazo
60 hits