Transcutaneous Pco2 to Monitor Noninvasive Mechanical Ventilation in Adults: Statistical Analysis

In: Mechanical Ventilation

17 Apr 2014

PaC02 and TcPco2 were correlated by linear regression, with calculation of Pearson’s coefficient of correlation (r).
We also calculated the bias and the limits of agreement between the parameters as described by Bland and Altman. The bias (d) is the mean difference between TcPco2 and PaC02 values; if s is the SD of d, the limits of agreement between the two methods are defined as follows: d±2s. Ninety-five percent of the values of (TcPco2 —PaC02) are expected to be within the limits of agreement.
Exponential fits for curves of PaC02 and TcPco2 in response to induced changes were computed using software (GraphPad “Prism” Software ; GraphPad Software; San Diego, Calif)-Analysis of variance for repeated measurements and Student’s t test were used for other comparisons. flovent inhaler

Results
Agreement between transcutaneous and arterial values for C02 was tested over a range of 26 to 71 mm Hg. TcPco2 as a function of PaC02 is shown on Figure 2. Both measurements were highly correlated (r=0.968, p<0.001), and linear regression was close to the identity line (TcPco2=1.116XPaC02—0.46). Figure 2 indicates that at high Pco2 values, there was a trend for TcPco2 to slightly overestimate PaC02. Figure 3 shows the d and limits of agreement between the TcPco2 and PaC02. The d was 0.75 mm Hg; SD of d was 2.6 mm Hg; limits of agreement were therefore as follows: —4.5 to +6 mm Hg. Eight of 26 patients were receiving low-dose dopamine (<0.2 mg/min). In these cases, correlation between TcPco2 and PaC02 was not altered and appeared as close as for the rest of the group (Fig 2).
In five patients with prolonged, continuous recording, the mean values for (TcPco2—PaC02) showed no significant difference at 0, 1, 2, 3, and 4 h (p=0.873). Therefore, no significant drift of TcPco2 could be detected over this period; maximal difference between TcPco2 and PaC02 reached 6 mm Hg after 4 h (Fig 4).

Figure 2. Regression line for TcPco2 on PaC02, and identity line; values are in mm Hg. Data shown are those of 26 patients without dopamine (open circles) or with low-dose dopamine (crosses).

Figure 2. Regression line for TcPco2 on PaC02, and identity line; values are in mm Hg. Data shown are those of 26 patients without dopamine (open circles) or with low-dose dopamine (crosses).

Figure 3. Bias of TcPco2 compared with PaC02 (d), and SD of bias (s). Values of (TcPco2 —PaC02) are plotted against mean value of TcPco2 and PaC02, as described by Bland and Altman. Data shown are those of 26 patients and are expressed in mm Hg. Patients are without dopamine (open circles) or with low-dose dopamine (crosses).

Figure 3. Bias of TcPco2 compared with PaC02 (d), and SD of bias (s). Values of (TcPco2 —PaC02) are plotted against mean value of TcPco2 and PaC02, as described by Bland and Altman. Data shown are those of 26 patients and are expressed in mm Hg. Patients are without dopamine (open circles) or with low-dose dopamine (crosses).

Figure 4. TcPco2 and PaC02 values for five patients over a 4-h period.

Figure 4. TcPco2 and PaC02 values for five patients over a 4-h period.


About this blog

Blog invites submissions of review articles, reports on clinical techniques, case reports, conference summaries, and articles of opinion pertinent to the control of pain and anxiety in dentistry.