OBSTETRIC ANESTHESIA Section Editor Mieczyslaw Finster

Epidural Test Dose and Intravascular Injection in Obstetrics: Sensitivity, Specificity, and Lowest Effective Dose Pietro Colonna-Romano, MD, Nagaraj Lingaraju, and Leonard E. Braitman, PhD

MD,

Steven D. Godfrey,

MD,

Department of Anesthesiology, Hahnemann University Hospital, Philadelphia, Pennsylvania

The authors studied the sensitivity and specificity of several epidural test doses as markers of intravascular injection in laboring patients in a prospective doubleblind, randomized study. Fifty-nine parturients were assigned randomly to receive an intravenous injection of either normal saline solution (3 mL, NS group) or 1.5% lidocaine with epinephrine 1:200,000 (1 mL, EPI-5 group; 2 mL, EPI-10 group; or 3 mL, EPI-15 group). The EPI-5 and EPI-10 doses were diluted to 3 mL volume with normal saline solution. All injections were given during uterine diastole. Maternal heart rate was monitored with a pulse oximeter. An observer who was unaware of the study treatment recorded the baseline and the peak maternal heart rate within the first minute after the injection and questioned the patient about tinnitus, dizziness, metallic taste, and palpitations. He then recorded his opinion as to whether the patient had received the saline or the test solution. Analysis of the maternal heart rate showed an average increase (baseline-topeak criterion) of 8 2 10 beatshin (mean k SD) in the NS group. In the other groups, the increase was 21 2 8 (EPI-5 group), 31.5 2 13 (EPI-10 group), and 29 2 9 beatsimin (EPI-15 group). A baseline-to-peak criterion of >10 beatdmin identified all intravascular injections in the EPI-15 (by design) and EPI-10 groups

T

he use of 15 pg of epinephrine in the epidural test dose (ETD) as a marker of intravascular injection was first described by Moore and Batra (1). Their study showed an increase in mean heart rate of approximately 30 beats/min in healthy nonpregnant volunteers after an intravenous injection of 15 pg of epinephrine. Since then, the use of epinephrine as a component of an ETD has been widespread. In 1990, Guinard et al. (2) determined the sensitivity of an epinephrine-containing ETD, as well as its lowest effective dose in healthy nonpregnant volunAccepted for publication April 10, 1992. Address correspondence to Dr. Colonna-Romano, Department of Anesthesiology, Hahnemann University Hospital, Broad and Vine Streets, Philadelphia, PA 19102-1192.

372

Anesth Analg 1992;75:3724

(15 of 15 and 14 of 14, respectively) with a sensitivity of 100%.Specificitywas 73% (11 of 15 true negatives). The calculated areas under the receiving operating characteristiccurves for the EPI-5, EPI-10, and EPI-15 groups were, respectively, 0.83 iz 0.8 (sE), 0.91 0.5, and 0.93 2 0.5. For the EPI-10 and EPI-15 groups, positive predictive value (+PV) ranged from 24% to 41% (assuming a prevalence between 8% and 16%). The negative predictive value (-PV) was 100%. The observer also correctly guessed the treatment of all patients in the EPI-10 and EPI-15 groups. Thus, the sensitivity was loo%, whereas the specificity reached 93% (14 of 15 true negatives). For the observer, the positive predictive value was 55%-73% and the negative predictive value 100%. Analysis of the symptoms (alone and in combination) exhibited low sensitivity (4 cm, uterine contractions at 53-4-min intervals) were enrolled in the study after they requested lumbar epidural analgesia. Their ASA physical status was I or I1 after an uncomplicated pregnancy. Patients with a history of hypertension of any etiology were excluded. The MHR was continuously monitored with a pulse oximeter (Nellcor, Hayward, Calif., model N-100, with a filtering time constant of approximately 7 s). A noninvasive blood pressure cuff was used to measure arterial blood pressure every 3 min. Fetal heart rate (FHR) was displayed continuously. Uterine activity was monitored with an external tocodynamometer or an internal uterine pressure catheter. Each patient received an intravenous infusion of 700-1000 mL of crystalloid solution through a peripheral catheter inserted into a vein in the dorsum of the hand. A lumbar epidural catheter was then inserted and taped in place. An observer (either an attending anesthesiologist or a resident with >6 mo in the anesthesiology program) who was unaware of the study treatment established a baseline MHR. This was defined as a stable rate with no variability for four to five consecutive beats in the absence of uterine contractions. The patients were then sequentially assigned to one of the following four groups: NS, EPI-5, EPI-10, or EPI-15. Fifteen patients received 3 mL of normal saline solution intravenously (NS group); other subjects were given an intravenous injection of 1.5% lidocaine with epinephrine 1:200,000. Specifically, 15 patients in the EPI-5 group received 1mL; 14 patients in the EPI-10 group received 2 mL; and 15 patients in the EPI-15 group received 3 mL of the test solution. The EPI-5 and EPI-10 test doses were diluted to 3 mL with normal saline solution. All injections were given in the absence of uterine contractions and immediately after the establishment of baseline MHR. Parturients who developed uterine contraction within 1min after the intravenous injection were also included in the study. An observer who was unaware of the study treatment then recorded the peak MHR, which was defined as the most rapid heart rate noted on the pulse oximeter within 1 min after the intravenous injection. The observer also questioned patients about the presence of tinnitus, dizziness, metallic taste, and palpitations and recorded their answers as well as his own opinion as to whether patients had received the test or the saline solution. No specific guidelines were given to the observer to discriminate among the groups, and the observer was not aware that different test doses were used. An obstetric nurse (blinded) and the authors (not blinded) observed the FHR tracing for signs of fetal distress. We used the Newrnan-Keuls test to compare the

OBSTETRIC ANESTHESIA COLONNA-ROMANO ET AL. SENSITIVITY AND SPECIFICITY OF ETD IN OBSTETRICS

373

Table 1. Peak Maternal Heart Rate Increases in the Four Study Groups Study group

Peak MHR (beats/min above baseline)

NS (n = 15) EM-5 (n = 15) EPI-10 (n = 14) EPI-15 (n = 15)

8 21 31.5 29

SD ? 10 28 ? 13

29

MHR, maternal heart rate; NS, normal saline solution (3 mL IV); EPI-5 (1 mL IV), EPI-10 (2 mL IV), EPI-15 (3 mL IV), 1.5% lidocaine with epinephrine 1:200,000. Data are means C SD; Newman-Keuls test: NS < EPIJ < EPI-10 = EPI-15 (I‘ < 0.05).

four peak MHRs and estimated the lower bound of the 95% confidence interval from the EPI-15 group (peak MHR) to discriminate among the groups. We then calculated the sensitivity and specificity of the increase in MHR (baseline-to-peak criterion) in the four groups. The areas under the receiving operating characteristic (ROC) curves for each study group assessed overall performance of this test (range 0.51.0). A ROC curve plots 1 - specificity on the horizontal axis against the sensitivity on the vertical axis at various possible cutoff points. Thus, in effect, a ROC curve displays the trade-offs between sensitivity and specificity that a diagnostic test allows. The ROC curves and the area under the ROC curves were determined with LABROC 1, a program for IBM-PCcompatible microcomputers based on maximal likelihood estimation of a binormal ROC curve. LABROC 1 uses a set of continuously distributed data. The positive predictive value (+PV) determined the diagnostic accuracy in the presence of a positive test (increase in MHR). The negative predictive value (-PV) determined the diagnostic accuracy in the presence of a negative test (no increase in MHR). Sensitivity and specificity of the observer’s clinical evaluation were also calculated, along with +PV and -PV. Finally, evaluation of the four symptoms as markers of intravascular injection of ETD was assessed by calculating specificity and sensitivity.

Results The results of the analysis of the peak MHR are shown in Table 1. The Newman-Keuls comparison test of the four peak MHRs showed NS < EPI-5 < EPI-10 = EPI-15 (P < 0.05). In the EPI-15 group, the average increase of the peak MHR was 29 5 9 beats/min (mean & SD) above the baseline. Using the lower bound of the 95% confidence interval for the EPI-15 group, we accepted an increase >10 beats/min above the baseline as diagnostic of intravascular injection of ETD. We then calculated sensitivity and specificity of the peak MHR in the four groups.

374

ANESTH ANALG 1992;75:3724

OBSTETRIC ANESTHESIA COLONNA-ROMANO ET AL. SENSITIVITY AND SPECIFICITY OF ETD IN OBSTETRICS

Table 2. Sensitivity and Specificity of the Peak Maternal Heart Rate and of the Observer Evaluation in the Four Studv Groups

2

0 c

Peak MHR

Observer evaluation

87% (13/15 TP) 100% (1414 TP) 100% (15/15 TP)

67% (10115 TP) 100% (1404 TP) 100% (15/15 TI’)

73% (11115 TN)

93% (la15 TN)

Study group

0 21 ] /

0

01

02

03

04

05

06

07

08

09

10

Sensitivity EPI-5 EPI-10 EPI-15 Specificity NS

FALSE POSITIVE FRACTION (1-SPECIFICITY)

Figure 1. ROC curve and area under the curve assessing the level of diagnostic accuracy in distinguishing 15 patients in the NS group from 14 patients in the EPI-10 group. Area = 0.9289; SEM = 0.0554. 1.0,

/

MHR,maternal heart rate; NS, normal saline solution (3 mL IV,n = 15); EPI-5 (1 mL IV, n = 15), EPI-10 (2 mL IV, n = 14), EPI-15 (3 mL IV, n = 15), 1.5% lidocaine with epinephrine 1:200,000; Sensitivity, number of true positives divided by the number of true positives plus false-negatives; Specificity, number of true negatives divided by the number of true negatives plus false-positives;TI’, true positive; TN, true negative.

Table 3. Sensitivity and Specificity of the Four Symptoms Attributable to the Systemic Effects of the Test Dose Symptom

U

c

::I, ,

0

01

02

Study group

,

,

,

,

03

04

05

06

07

~

08

,

,

09

10

FALSE POSITIVEFRACTION (1 -SPECIFICITY)

Figure 2. ROC curve and area under the curve assessing the level of diagnostic accuracy in distinguishing 15 patients in the NS group from 15 patients in the EPI-15 group. Area = 0.9107; SEM = 0.0539.

The baseline-to-peak criterion identified all the intravascular injections in the EPI-15 group by design (the lower bound of the 95% confidence interval was calculated from the EPI-15 group) and in the EPI-10 group with a sensitivity of 100% and a -PV of 100%. Specificity of this test was 73% (11 of 15 true negatives), and the +PV ranged between 24% and 41%, assuming a prevalence of unintentional intravenous cannulation between 8% and 16%. The areas under the ROC curves for the EPI-10 and EPI-15 groups are displayed, respectively, in Figures 1and 2. They were 0.83 for the EPI-5 group, 0.91 for the EPI-10 group, and 0.93 for the EPI-15 group. The observers were also abIe to idenbfy all the intravascular injections of ETD in the EPI-15 and EPI-10 groups with a sensitivity of 100%. Based on clinical evaluation, specificity reached 93% (14 of 15 true negatives). Results are shown in Table 2. The +PV was 55%-73% (prevalence 8%-16%), and the -PV 100%. Sensitivity and specificity of the four symptoms, attributable to the systemic effects of an intravascular

Dizziness

Palpitations

Tinnitus

1/15 3/14 2/15

4/15 6/14 5115

4/15 3/14 7115

0115 0114 4/15

13/15

13/15

14/15

15/15

~~

~

,

Metallic taste

Sensitivity EPI-5 EPI-10 EPI-15” Specificity

NS

NS, normal saline solution (3 mL IV, n = 15); EPI-5 (1 mL IV, n = 15), EPI-10 (2 mL IV, n = 14), EPI-15 (3 mL IV, n = 15), 1.5% lidocaine with epinephrine 1:200,000. Sensitivity, number of true positives divided by the number of true positives plus false-negatives;Specificity, number of true negatives divided by the number of true negatives plus false-positives. Sensitivity of the symptoms analyzed in combination reached 67%in the EPI-15 POUP.

injection of ETD, were also calculated and did not reach clinical significance. The results are shown in Table 3. The FHR monitor did not show any tracings indicative of fetal distress, and no medical intervention of any kind was ever required during the performance of the study. Noninvasive arterial blood pressure measurements provided frequently spurious results (movement artifacts), and the data were not used in the study.

Discussion The intravenous injection of toxic doses of local anesthetics may cause seizures, cardiac arrest, and death (4). To prevent this occurrence during the performance of an epidural block, the concept of an “epidural test dose” (ETD) was developed. An ETD should reliably provide an indication of

ANESTH ANALG 1992;75:372-6

unintentional intravenous insertion of an epidural catheter. The frequency of this mishap has been reported to be in the range of 3%-5% in the general population (5) and between 8% and 16% in pregnant patients (6), probably due to dilated epidural vessels (7). Several techniques and agents have been proposed for use as ETDs. Since Moore and Batra (1) demonstrated a predictable heart rate increase after an intravenous injection of epinephrine (15 pg) in nonpregnant volunteers, the use of this agent as a marker for intravascular injection has become common. Recently, it was shown that the same heart rate changes could be induced with intravenous epinephrine (10 but not 5 pg) in nonpregnant volunteers (2). In laboring patients, the sensitivity and specificity of epinephrine as a marker for intravascular injection are low (3) because of the wide variability in MHR. Isoproterenol has been suggested to replace epinephrine (8) as a component of an ETD in obstetrics; however, healthy and preeclamptic parturients have different chronotropic responses to the drug (9). The sensitivity of ephedrine is poor (lo beats/min above the baseline) in the NS group 75% of the time (4 of 15 false-positive by the baseline-to-peak criterion vs 1 of 15 false-positive based on the observer’s evaluation). In practical terms, a +PV of 55%-73% implies that a positive test may yield false-positive results 45%-27% of the time, leading to unnecessary replacement of the epidural catheter. Finally, in our study 10 p g of epinephrine was as effective as 15 pg in detecting intravascular injection. The epinephrine-induced reduction in uterine blood flow in animals has been shown to be dose dependent (14); therefore the use of 10 p g in humans may be warranted. In conclusion, in this study we found that (a) epinephrine is a sensitive marker of intravascular injection in laboring patients; (b) within 1 min after injection, the baseline-to-peak criterion was a reliable indicator of intravenous injection; (c) 10 p g of epinephrine is as effective as 15 p g ; (d) symptoms attributable to the systemic effects of our ETD were neither sensitive nor specific; (e) the pattern of epinephrine-induced tachycardia could be clinically differentiated from tachycardia from other sources; and (f) on the basis of our observer’s evaluation, between 27% and 45% of epidural catheters may be

unnecessarily repositioned in the presence of a positive test. We thank J. C. Horrow, MU, for his valuable suggestions concerning the preparation of this manuscript.

References 1. Moore DC, Batra MS. The components of an effective test dose

prior to epidural block. Anesthesiology 1981;55:693-6. 2. Guinard JP, Mulroy MF, Carpenter RL, Knopes KD. Test dose: optimal epinephrine content with and without acute betaadrenergic blockade. Anesthesiology 1990;73:386-92. 3. Leighton BL, Norris MC, Sosis M, Epstein R, Chayen B, Larijani GE. Limitations of epinephrine as a marker of intravascular injection in laboring women. Anesthesiology 1987;66: 688-91. 4. Marx GF. Cardiotoxicity of local anestheticsthe plot thickens (editorial). Anesthesiology 1984;60:%5. 5 . Dawkins CJM. Analysis of the complications of extradural and caudal block. Anaesthesia 1969;24:554.63. 6. Ahn NN, Ung DA, DeFay S, Cannelli G, Rudy TE. Blood vessel puncture with epidural catheters (abstract). Anesthesiology 1989;71:A916. 7. Kennepp NB, Gutsche BB. Inadvertent intravascular injections during lumbar epidural anesthesia (letter). Anesthesiology 1981;54:172-3. 8. Leighton BL, DeSimone CA, Norris MC, Chayen B. Isoproterenol is an effective marker of intravenous injection in laboring women. Anesthesiology 1989;70:206-9. 9. DeSimone CA, Leighton BL, Norris MC, Chayen B, Menduke H. The chronotropic effect of isoproterenol is reduced in term pregnant women. Anesthesiology 1988;69:6264. 10. Fong J, Gadalla F, Fiamengo SA, Druzin M, Artusio JF. Ephedrine sulfate as the intravenous component of the epidural analgesia test dose in laboring parturients (abstract). Anesthesiology 1988;69:A706. 11. Leighton BL, Gross JB. Air: an effective indicator of intravenously located epidural catheters. Anesthesiology 1989;71:84851. 12. Chestnut DH, Owen CL, Brown CK, Vandewalker GE, Weiner CP. Does labor affect the variability of maternal heart rate during induction of epidural anesthesia? Anesthesiology 1988; 68:622-5. 13. Cartwright PD, McCarroll SM, Antzaka C . Maternal heart rate changes with a plain epidural test dose. Anesthesiology 1986; 65:22&8. 14. Hood DD, Dewan DM, James FM. Maternal and fetal effect of epinephrine in gravid ewes. Anesthesiology 1986;64:610-3. 15 Shnider SM, Levinson G. Anesthesia for obstetrics. 2nd ed. Baltimore: Williams & Wilkins, 198775-6. 16 van Zundert A, Vaes L, Soetens M, et al. Every dose given in epidural analgesia for vaginal delivery can be a test dose. Anesthesiology 1987;67436-40. 17 Roetman KJ, Eisenach JC. Evaluation of lidocaine as an intravenous test dose for epidural anesthesia (abstract). Anesthesiology 1988;69:A669. 18 Grice SC, Eisenach JC, Dewan DM, Mandell G. Evaluation of 2-chloroprocaine as an effective intravenous test dose for epidural anesthesia (abstract). Anesthesiology 1987;67A627. 19 Gissen AJ, Datta S, Lambert D. The chloroprocaine controversy: is chloroprocaine neurotoxic? Reg Anaesth 1984;9:13.!% 45. 20 Lingaraju N, Colonna-Romano P, Mazala M. Lidocaine 45 mg in the epidural test dose is not a marker of intravascular injection in parturients (abstract). Anesthesiology 1990;73: A976.