Correlation of CO2 Measured by Blood Gas vs Transcutaneous Monitor

Background and Significance/Preliminary Studies Carbon dioxide (CO2) measurement is part of
the fundamental respiratory evaluation in an ICU, as both high and low values of CO2 can have
detrimental effects on neonatal health. The partial pressure of CO2 reflects ventilation,
which is the elimination of carbon dioxide. The ABG sample is the clinical tool most commonly
used to evaluate acid-base physiology at the bedside. It directly measures the pH, arterial
partial pressure of oxygen, and arterial partial pressure of CO2. Though direct measurement
of CO2 in the arterial blood is the most accurate way to assess the amount of CO2, it
requires blood sampling from the arteries and does not provide continuous monitoring of CO2.

A non-invasive tool for measuring PaCO2, such as the SenTec monitor, can help decrease blood
draws, thereby decreasing the incidence of iatrogenic anemia and the need for neonatal blood
transfusions. Blood samples are collected through arterial lines, capillary heel sticks, or
peripheral arterial punctures which are invasive and can be painful for the patient and may
potentially result in injury. Although an ABG in neonates typically only requires 0.3mL-0.5mL
of arterial blood, the need for frequent ABG sampling (every two to six hours) may result in
a significant amount of iatrogenic blood loss. A newborn has a total blood volume of only
80mL/kg. In addition, the transcutaneous sensor has potential monitoring advantages as
measurements are continuous rather than representative of a single point in time, as seen in
ABGs. Continuous monitoring can help to avoid fluctuations in CO2 levels which can
potentially result in neurologic injury in preterm babies, such as intraventricular
hemorrhage and periventricular leukomalacia. One of the most beneficial aspects of the SenTec
monitor is that it displays CO2 levels in real time, allowing clinicians to monitor trends
and to intervene sooner if needed.

Previous studies have been conducted on transcutaneous monitors in newborns. One such study
evaluated the CO-OXSYS Digital Sensor (also known as the V-Sign Digital Sensor) made by
SenTec. 25 patients between the ages of 10 months to 79 years of age were studied. Each
sensor was placed on the tragus of each patient's ear. An ABG was obtained fifteen minutes
after sensor application and TCCO2 measurements were recorded at the same time. ABG samples
were analyzed with a Rapidpoint 405 analyzer. A reliable correlation was found between the
TCCO2 and the PaCO2 readings (R2-0.61, p=<0.001). It was noted that age and temperature did
not affect the accuracy of the CO2 readings. This study concluded that with more sampling,
the CO-OXSYS Digital Sensor would be useful in a variety of clinical settings.

A second study was performed assessing the SenTec monitor during neonatal High Frequency
Oscillatory Ventilation. Fourteen neonatal patients were studied. The digital sensor was
applied to each patient on varying body parts that included the left chest, right chest, and
liver. Fifteen minutes after sensor application, an ABG was obtained and the simultaneously
displayed TCCO2 measurement was recorded. ABG measurements were subsequently analyzed using a
Siemens Rapidpoint 405 analyzer. A clinically acceptable correlation was found between the
TCCO2 and PaCO2 readings (R2=0.807, p=<0.001). This study suggested the monitor could
certainly be used as an alternative to ABGs.

In 2004 a study reviewed the TOSCA monitor, made by Linde Medical Sensors, which combined
pulse oximetry and TCCO2 monitoring. This machine used a single ear sensor which worked at 42
degrees C to enhance blood flow in capillaries below the sensor. This study included 60
neonates who had an ear sensor placed on the right earlobe. Ten minutes after placement, an
ABG was performed and compared to the TCCO2 reading. The transcutaneous readings were found
to be clinically acceptable when compared to the ABGs. A secondary benefit of the
transcutaneous monitoring on the earlobe was a decrease in the head movement of the neonate
to aid in maintaining a patent airway. In addition, the sensor did not have to be removed for
chest radiographs or while the parent was bonding with his or her infant in their lap (also
known as "kangaroo care").

A recently completed study used the actual SenTec monitor I will be using. A total of 15
patients were included in this study with the following inclusion criteria:

Current weight > =1000grams Need for mechanical ventilation Arterial access Signed informed
consent by parents/legal guardians The sensor was placed on the abdominal wall and when a
routine ABG sample was obtained, the reading from the transcutaneous device was recorded.
Photographs were also taken throughout the study to evaluate any injury to the skin from the
monitor. The photographs were randomized and a blinded reviewer rated the photographs as
normal, mild erythema, or burn/blistering. The reviewer of the photographs confirmed that
there were no adverse skin changes. It was concluded that the TCCO2 measurements obtained
correlated with the PaCO2 values.

Study Aims This research study is designed to assess the accuracy of SenTec transcutaneous
monitors for CO2 monitoring in neonatal patients. The data from this research may be used to
implement the SenTec monitors as part of the regular monitoring equipment used in the NICU at
Memorial University Medical Center. These results will contribute to my course of study by
aiding in my future research paper publication and poster presentation (to other ICU
attendings).

Administrative Organization The Level III NICU at Memorial University Medical Center will be
the sole participating unit. Registered nurses and/or respiratory therapists within the NICU
will be collecting the routine ABG samples. Respiratory therapists within the NICU will
analyze the ABG samples.

Study Design The design of the study will be an open-label, correlational prospective study.
The study population will be any neonate weighing greater than one kg at birth requiring an
arterial catheter for monitoring for cardiorespiratory distress. At the end of my study, if
it is found that the SenTec digital monitor is as accurate and valid as the ABG samples
collected and that no adverse skin changes are observed, I hope that the SenTec monitoring
system will be implemented into Memorial's regular NICU monitoring equipment. With approval
from the SenTec manufacturer, I will aim to write a paper and submit it to an appropriate
neonatal or critical care journal for future publication. In addition to the paper, I plan to
create a poster for presentation to critical care (PICU and NICU) unit staff (physicians,
RTs, RNs, and NPs) at Memorial University Medical Center.

Study Procedures The sampling plan will include any neonate with a birth weight over 1000g,
requiring an arterial line. Patients with significant pre-existing skin breakdown over the
abdominal area will be excluded from the study. Recruitment will occur exclusively in the
NICU. Consent will be obtained in the NICU once the arterial line has been placed. As the
Principal Investigator, I, and/or other Dr. Daniel Sandler will obtain consent. We will
provide the patient's parents with a summary of the protocol of the study and brochures
providing information about the sensor and its manufacturer.

There will be no placebo study agents. This is going to be a validation study. The sensor
will be stored in the NICU. The sensor will be placed on the patient by the patient's bedside
RN. Potential adverse effects should be minor and limited to mild, local skin reactions.

Serial TCCO2 measurements will be compared to PaCO2 levels measured on arterial blood gas
(ABG) samples obtained simultaneously on these patients. The sensor will be calibrated and
applied to the skin according to manufacturer instructions. Whenever an arterial blood gas is
collected for clinical indications, the transcutaneous monitor's reading will be
simultaneously recorded in a data log including the patient's initial diagnosis, gestational
age, sex, sensor location, birth weight, and body temperature. The skin of each infant will
be closely inspected for injury (such as irritation or breakdown) before the skin probe is
applied, and after each time that the probe is removed. Application skin sites will also be
photographed on entry to and exit from the study, and whenever the probe is removed.

The photographs taken and the data log will be stored in an Excel document on a Memorial
University Medical Center computer in the NICU. All study information will be password
protected. Only the investigators (Charli Cohen, Dr. Daniel Sandler), and Eric Clayton, the
study statistician, will have access to the study data collected.

Safety Monitoring Plan There is a slight risk for minor skin reactions from the probe. This
sensor has been tested in newborns and it seems that any adverse skin events have been
minimal. The probe has not yet been tested in extremely premature babies (weighing less than
one kilogram at birth). We will constantly be monitoring the patient's safety during routine
unit checks. The study investigators, with the assistance of NICU staff members, will
identify, document, and report adverse events. A patient will be removed from the study if
the skin reaction becomes severe. To carefully monitor for skin changes, photographs will be
taken before and after application of the probe.

Analysis Plan

Hypotheses:

H0: Observed mean difference between PaCO2 and TCCO2 < 10 mmHg. Ha: Observed mean difference
between PaCO2 and TCCO2 > 10 mmHg.

The study data will be analyzed using the Bland Altman technique for assessing agreement
between two methods of clinical measurement6. Consistency between the PaCO2 and TCCO2 methods
will be quantified by constructing limits of agreement (LoA). These limits are determined by
using the observed mean and standard deviations of the differences between the two methods. A
scatterplot will be constructed in which the Y-axis contains the differences in the paired
measurements for each case (TCCO2 - PaCO2) and the X-axis contains the means of the same
paired measurements ((TCCO2 + PaCO2)/2). A line representing the observed mean difference
will be constructed parallel to the X-axis, as will two more lines at the points ± 1.96SD of
the mean difference. These lines represent the LoA. 95% confidence intervals for the upper
and lower LoA will be calculated. The pre-defined maximum acceptable difference between the
methods of measurement has been determined as 10 mmHg. If the upper 95% confidence interval
of the upper LoA and the lower 95% confidence interval of the lower LoA are both less than 10
mmHg from the observed mean difference, the H_0 will be accepted and the two measurement
methods will be said to be in agreement.

Sample Size:

Using the published data and sample size methods, with α=0.05, β=0.20, expected mean
difference (D = -3.21 mmHg), expected standard deviation (SD = 3.01 mmHg), and the clinically
significant maximum allowed difference between the methods (δ=10 mmHg), the required minimum
number of paired measurements is 266.

Inclusion Criteria:

- Neonates in NICU with an arterial line

Exclusion Criteria:

- Birth Weight < 1000g