Introduction The role of transcutaneous carbon dioxide monitoring (P_tcCO₂) for patients with respiratory failure has been studied in a variety of clinical settings. However, its accuracy compared to arterial partial pressure of carbon dioxide (PaCO­₂) in patients undergoing non-invasive ventilation (NIV) for hypercapnoeic respiratory failure has not been validated. The degree to which PtcCO₂ approximates ‘gold-standard’ PaCO­₂ in this context was evaluated in a meta-analysis including 16 prospective observational studies.

Methods 16 prospective studies evaluating PtcCO₂ as a correlate for PaCO₂ in patient cohorts undergoing NIV were included. In all cases, Bland-Altman analysis was used to compare agreement among measures. Mean bias between the two methods of measurement was calculated for each study, along with standard deviations for each data set. The mean weighted bias was subsequently calculated using the formula (Bias ⋉ N)/N

Results Mean weighted bias for the populations studied was 0.06 kPa ( ± 2SD 2.40 kPa).

Discussion Patients with hypercapnoeic respiratory failure undergoing non-invasive ventilation (NIV) require regular measurements of PCO₂ and pH status in order to optimise treatment. The current gold standard for measurement of PCO₂ is arterial blood gas sampling. In critical care environments this necessitates placement of an indwelling arterial catheter, a painful procedure which carries a risk of complications including thrombotic occlusion, distal ischaemia, infection and pseudoaneurysm formation. Whilst accurate, arterial PCO₂ measurement provides a single, static reading of an inherently dynamic process in the context of hypercapnoeic respiratory failure.

Transcutaenous PCO₂ monitoring may prove a suitable and reliable non-invasive method for analysing partial pressure of carbon dioxide, with good approximation to arterial values. Our results suggest that transcutaneous CO₂ monitoring may be used as a reliable surrogate to arterial CO₂ measurements in specific patient populations undergoing NIV. However, variance is wide and further studies are needed to establish whether PtcCO₂ is a sufficiently accurate marker on which to base treatment decisions.