Analysis of Physiological Respiratory Variable Alarm Alerts Among Laboring Women Receiving Remifentanil

Remifentanil may be used by laboring women for analgesia, despite controversy because of potential apneas. We evaluated candidate variables as early warning alerts for apnea, based on prevalence, positive predictive rate, sensitivity for apnea event detection, and early warning alert time intervals (lead time) for apnea.

We performed a secondary analysis of respiratory physiological data that had been collected during a prospective IRB-approved study of laboring women receiving IV patient-controlled boluses of remifentanil 20 to 60 μg every 1 to 2 minutes. Analyzed data included the respiratory rate (RR), end-tidal CO2 (EtCO2), pulse oximetry (SpO2), heart rate (HR), and the Integrated Pulmonary Index (IPI; Capnostream 20; Medtronic, Boulder, CO) that had been recorded continuously throughout labor. We defined immediate early warning alerts as any drop in a variable value below a prespecified threshold for 15 seconds: RR < 8 breaths per minute (bpm), EtCO2 < 15 mm Hg, and SpO2 < 92%. We defined alerts as "sustained" when the value remained below the threshold for ≥ 10 further seconds. The IPI value (1 to 10; 10 = healthy patient, ≤4 = immediate attention required, 1 = dire condition) was generated from a proprietary algorithm using RR, EtCO2, SpO2, and HR parameters. Apnea was defined as maximal CO2 < 5 mm Hg for at least 30 consecutive seconds.

We counted 62 apneas, among 10 of 19 (52.6%) women who received remifentanil (total dose 1725 ± 1392 μg, administered over 160 ± 132 minutes). We counted 331 immediate early warning alerts for the variables; 271 (82%) alerts were sustained for ≥10 seconds. The positive predictive value of alerts for apnea was 35.8% (99% confidence interval [CI]: 27.1-45.6), 28.9% (99% CI: 20.8-38.7), 4.3% (99% CI: 1.9-9.6), and 24.6% (99% CI: 18.3-32.2) for RR, EtCO2, SpO2, and IPI, respectively. The sensitivity for apnea event detection was 100% (99% CI: 90.3-100) for RR (<8 bpm) and IPI (≤4); 75.8% (99% CI: 59.8-86.9) for EtCO2 <15 mm Hg; and 14.5% (99% CI: 6.5-29.4) for SpO2 <92%. We found a statistically significant difference in the timing of RR, EtCO2, SpO2, and IPI alerts for apnea; Friedman's Q = 33.53; P < .0001. The EtCO2 had a median (interquartile range) lead time of -0.2 (-12.2 to 0.7) seconds, and SpO2 had a median (interquartile range) lead time of 40.0 (40.0 to 40.0) seconds.

The majority of women receiving IV remifentanil for labor analgesia experienced apneas. Alerts for EtCO2 (<15 mm Hg), RR (<8 bpm), and IPI (≤4) detected most apneas, whereas SpO2 alerts missed the majority of apneas. All variables had a low positive predictive rate, demonstrating the limitations of the respiratory monitors utilized as early warning surveillance for apneas in this setting.