Berechnung von CO2-Verlaufswerten aus Pulsoximetrien von Patienten unter nichtinvasiver nächtlicher Beatmung

Respiratory insufficiency can be differentiated into hypoxemic and hypercapnic respiratory insufficiency. In both forms, an acute and a chronic course can be distinguished. The most common causes of chronic hypercapnic/ventilatory insufficiency are diseases in which exhaustion of the inspiratory respiratory muscles occurs. These include neuromuscular disease, kyphoscoliosis, obesity hypoventilation syndrome, and chronic obstructive pulmonary disease. As the disease progresses, hypercapnia occurs at night as well as during the day. Patients notice this primarily by the appearance of exertional dyspnea. However, the risk of the disease is worsening with increasing pCO2 levels and CO2 narcosis. To prevent this, non-invasive ventilation is used in addition to invasive ventilation. The indication for noninvasive ventilation is primarily based on the degree of hypercapnia, i.e., the measured pCO2 value. The pCO2 can be determined using various methods. While end-tidal measurement does not play a role in sleep medicine, transcutaneous measurement and determination from arterial or capillary blood by means of a blood gas analysis are used as standard. However, continuous pCO2 measurement is required for monitoring and optimal adjustment of noninvasive ventilation. The only method currently available for this purpose is transcutaneous pCO2 measurement. According to several studies, it provides clinically acceptable values compared to blood gas analysis and can therefore be considered the gold standard of pCO2 progression measurement. However, the maintenance of transcutaneous measurement devices is very expensive, so the aim of this study was to validate a more cost-effective alternative method. This study is the first to calculate pCO2 history values from pulse oximetry of patients under noninvasive nocturnal ventilation after one-point calibration using a patented formula. For this purpose, 20 patients under noninvasive ventilation, 10 of whom received supplemental oxygen, were recruited. Nocturnal transcutaneous pCO2 was measured with TOSCA 500 and at least one capillary blood gas analysis was taken for one-point calibration. Following the measurement, the nocturnal pCO2 trend was calculated from the collected data for each patient using the basic formula and an SpO2-corrected formula and compared with the transcutaneous readings. The aim of the SpO2-correction was to better reflect pCO2 increases during nocturnal oxygen desaturations or leaks. The graphical representation of the calculated and transcutaneously measured pCO2 values indicated a similar curve shape, which could be further improved by the SpO2-correction. Except for four patients, the mean deviation of the two methods was less than ± 4 mmHg in both the individual and group evaluations. Regression analysis demonstrated a strong positive and statistically significant correlation for the patients with noninvasive ventilation without oxygen therapy, those with noninvasive ventilation and oxygen therapy, and also for the patients with severe nocturnal pCO2 fluctuations ≥ 10 mmHg. The correlation increased even more with the help of SpO2-correction compared with the basic formula. Bland-Altman analyses revealed a bias for the different groups ranging from -0,84 mmHg to 0,36 mmHg. The limits of agreement ranged from a maximum of -8,16 mmHg to 8,33 mmHg. In summary, it has been shown for the first time that CO2 history values can be calculated reliably and clinically acceptable from pulse oximetry of patients under noninvasive nocturnal ventilation. However, in order to achieve the overall goal of developing a device that can calculate the pCO2 progression values directly from pulse oximetry after one-time calibration using a blood gas analysis, studies with a larger sample are necessary. In this context, the factor calculation of the SpO2-correction should be further adapted and the hemoglobin value of the patients should be tested as another component of the calculation formula. In addition, the standardized use of the pulse frequency average for calibration and an automated inclusion of the phase shift must be investigated.