Auscultated fetal heart rate accelerations

Auscultated fetal heart rate accelerations III. Use of vibratory acoustic stimulation Lisa L. Paine, CNM, MS, MPH,"·b Timothy R. B. Johnson, MD," and Greg R. Alexander, MPH, ScDb Baltimore, Maryland The auscultated acceleration test has been proposed as a simple, inexpensive screening test for fetal health; previous studies of the auscultated acceleration test used external stimulation to elicit fetal movement. This study was conducted to explore the ability of the auscultated acceleration test to predict nonstress test results when vibratory acoustic stimulation is used to elicit fetal reactivity. After antepartum nonstress testing on 100 gravid women between 28 and 43 weeks' gestation, a 6-minute auscultated acceleration test protocol was performed with two vibratory acoustic stimulations to the maternal abdomen if no spontaneous fetal heart rate acceleration occurred. The ability of the auscultated acceleration test to predict nonstress test results after selected variables were controlled for was as follows: sensitivity, 75%; specificity, 97.6%; false-positive results, 14.3%; and false-negative results, 4.7%. Logistic regression analysis indicated that, in addition to the auscultated acceleration test, gestational age and race contributed significantly to the prediction of nonstress test results. Although specificity and the false-positive rate were improved, the use of vibratory acoustic stimulations to elicit fetal movement did not improve the validity of the auscultated acceleration test in terms of sensitivity and false-negative results over previous studies. However, the auscultated acceleration test continues to show potential as an initial screening test for fetal assessment. In addition to recommendations for further research, methodologic issues related to sampling techniques are identified. (AM J OBSTET GYNECOL 1988;159:1163-7.)

Key words: Auscultated fetal heart rate, vibratory acoustic stimulation, antepartum nonstress test, research methodology

Observing accelerations of the fetal heart rate ( FHR) is a well-documented method of fetal assessment; the most commonly used method is the electronically monitored nonstress test (NST). 1· 3 In part because of the technologic contraints involved in antepartum fetal testing, only women with risk factors for poor perinatal outcomes are routinelv tested. Auscultating FHR accelerations has recently been explored as a simple, inexpensive screening test for fetal health in low- and high-risk pregnancies. 0 In light of mounting criticism toward introducing electronic fetal monitoring techniques into developing countries, 8 an auscultated acceleration test, predictive of NST results, could potentially serve as a primary assessment technique in areas of the world where electronic monitoring technology is unavailable. Vibratory acoustic stimulation has been extensively From the Fetal Assessment Center, Division of Maternal-Fetal Medicine, Department of G_vnecology and Obstetrics, The Johns Hopkins University School of Medicine," and the Department of Maternal and Child Health, The Johns Hopkins University School of Hygiene and Public Health-" Receivedfor publication January 13, 1988; revised April 23, 1988; accepted Ma_v 22, 1988. Reprint requests: Lisa L. Paine, CNM, MS, MPH, Department of GYNIOB, HOUCK 228,Johns Hopkins Hospital, 600 N. Wolfe St., Baltimore, MD 21205.

explored as a means of reducing both NST testing time and the incidence of nonreactive results. 9 · 10 Previous work on the auscultated acceleration test involved external stimulation of the fetus to elicit fetal movement. 7 This study explores the ability of the auscultated acceleration test to predict NST results when vibratory acoustic stimulation is used to stimulate fetal reactivity.

Material and methods After an initial antepartum NST, an auscultated acceleration test was performed on 100 women at the Johns Hopkins Hospital Fetal Assessment Center between July 1, 1987, and October 22, 1987. Gravid women from 28 to 43 weeks' gestation with a single fetus were given a thorough explanation of the procedures and asked to consent to participation. The NST was performed by the Fetal Assessment Center nurse and interpreted by an attending perinatologist. The usual criteria of at least two FHR accelerations of at least 15 beats/min, lasting 15 seconds within a 20minute period, were used to determine NST reactivity. Whenever possible, fetal monitoring was continued throughout the auscultated acceleration test to examine validity of the auscultated accelerations. Blinded to NST results and to simultaneous record1163

1164

Paine, Johnson, and Alexander ..\111

•= t=

FHR

'.\member llJ88 ( ;, 11crnl

.J ( )hstct

FHR

FM

AS "" AcoU1tlc Stlm.

MIN 5SEC. 1 1----+--,-,---r-t---t-+-ii---t--+-i-t-r--t-'-1~0-H --t-H~--t--r-r-rr-r-rTT-r.-r-rr-r-r.-r-r,-,rl 240 20 ~~-H-+-r::r __ =r::--:: J: 4- [;,.f.;,lj::_L-__j__~~r++-++-+-++-++++-lr++-++-+-+-1 228 19

216

18

,4

17

192

16 15 14

?

180 168

'"

144 132 120 108 96 84

t

:iX ~ =-:1 ~ =-_r1-1-+--1-++--+-++--1-++--1-1--+-11-++--+-+-+-1 I

120'

'.

12 11 10 09 08 07 Seconds

Fig. 1. Auscultated FHR acceleration and simultaneous electronic fetal monitor tracing with fetal movement (FM) after vibratory acoustic stimulation.

related variables. A standard statistical computer package was used for this analysis.

Table I. Most frequent indications for antepartum NST n

Intrauterine growth retardation Postdatism Decreased fetal movement Pregnancy-induced hypertension Gestational diabetes Postamniocentesis Previous pregnancy loss Class B or greater diabetes mellitus Chronic hypertension

21 14 10 9 8 8 8 4 3

ing of the FHR, one author (L. L. P.) performed the auscultated acceleration test using a previously reported technique.' The FHR was auscultated for 2 minutes with an Allen fetoscope and recorded on an auscultated acceleration test graph specifically designed for recording auscultated FHR accelerations; the baseline FHR was established during this time. Only women with a baseline FHR between 102 and 160 beats/min were included in this study. If an acceptable FHR acceleration (a rise of at least 2 beats per 5-second period over the baseline) was not auscultated in the first 2 minutes, a 2-second vibratory acoustic stimulation was applied to the maternal abdomen in the vicinity of the fetal vertex with a Western Electric model 5C artificial larynx. If an FHR acceleration was not auscultated in the next 2-minute auscultation period, vibratory acoustic stimulation was applied a second time in the same manner. If a single acceptable FHR acceleration occurred at any time during the 6 minutes of auscultation, the auscultated acceleration test was considered reactive (negative). If no FHR acceleration occurred or ifthe fetus did not move, the auscultated acceleration test was considered nonreactive (positive). Logistic regression analysis was used to assess the predictive relationship of the auscultated acceleration test with NST results, taking into account potentially

Results

The mean age of the women was 25.6 years; 19 were from 14 to 19 years old, 54 were 20 to 29 years old, and 27 were 30 to 44 years old. NSTs were performed on 80 black women and 20 white women. Nulliparous women represented 32 of those tested and 68 were multiparous. The mean gestational a'ge at the time of testing was 36.2 weeks; 19 were tested during the 28to 32-week period, 26 during the 33- to 36-week period, 41 during the 37- to 40-week period, and 14 during the 41- to 43-week period. Vibratory acoustic stimulation was performed once in 62 cases and twice in 13 cases; it was not required in 25 cases. The most frequent indications for testing are given in Table I. In order to validate accurate auscultation of accelerations, the electronic fetal monitoring strips obtained during the auscultated acceleration test were evaluated. Simultaneous tracings were obtained for 72 of the 100 auscultated acceleration tests performed. In 60 of the 63 monitoring strips indicating an FHR acceleration, an acceleration was accurately auscultated. In seven of the nine monitoring strips indicating no acceleration, the absence of FHR acceleration was correctly documented. The percentage agreement for the simultaneously monitored and auscultated FHR accelerations was 89.3. The kappa statistic, which calculates agreement between two measures after agreements that could be expected to arise by chance alone are taken into consideration, was 0.77. When kappa statistic results are evaluated, >0. 75 indicates excellent agreement beyond what would have occurred by chance alone; 0.40 to 0.75 indicates fair-good agreement; and <0.40 indicates poor agreement beyond chance. 11 Fig. 1 illustrates the similarity between the auscultated FHR accelerations and simultaneous electronically monitored accelerations during fetal movement stimulated by vibratory acoustic stimulation. Table II provides a summary of the results of the

Auscultated acceleration test with vibratory acoustic stimulation

Volume IS!J 1\umber 5

1165

Table II. Logistic regression results of prediction of NST by auscultated acceleration test and related variables p NSTs Initial NSTs (n = 100*)

All NSTs (n = 154*)

Significant variable Gestational age (wk) Auscultated acceleration test Black race Gestational age (wk) Auscultated acceleration test Previous pregnancy loss

Beta

Odds

Value

R

- 0.5917 3.0535

1.8 21.2

0.0002 0.0037

-0.373 0.270

3.4659

32.0

0.0496

0.145

-0.5697 2.5493

1.8 12.8

0.0000 0.0009

-0.389 0.264

3.0653

21.4

0.0009

0.264

*NSTs performed on 100 gravid women.

Table III. Summary classification of predictive value of auscultated acceleration test on NST based on logistic regression n = 100 Sensitivity (%) Specificity (%) False-positive rate (%) False-negative rate (%) Positive predictive value (%) Negative predictive value (%) Kappa statistic Agreement(%) Prevalence of nonreactive NST (%)

stepwise logistic regression analysis of the 100 cases studied. Variables considered in the logistic regression analysis that did not meet the 0.50 significance level were age, parity, and all of the indications for testing listed in Table I. The statistically significant variables in the model were gestational age (p < 0.001), the auscultated acceleration test (p < 0.005), and race (p < 0.05) (Table II). These findings indicate that the auscultated acceleration test is predictive of NST results. In addition, women tested at earlier gestational ages and black women were significantly more likely to have a nonreactive NST. Table III includes a summary of the classification table generated as a result of this analysis. The predictive ability of the auscultated acceleration test on the NST when the other variables in the model are taken into account was as follows: sensitivity, 75.0%; specificity, 97.6%; false-positive rate, 14.3%; and false-negative rate, 4.7%. The kappa statistic for agreement was 0.48, indicating good agreement in NST -auscultated acceleration test results. In the data reported above, each women studied contributed only one NST -auscultated acceleration test pair to the analysis. Tests, rather than patients, have been used as the unit of analysis in a number of previously published studies to evaluate antepartum testing techniques. 2 · 6 · 12 This approach raises methodologic issues of bias regarding the independence of the ob-

75.0 97.6 14.3 4.7 85.7 95.3 .48 94.0 16.0

n

=

154

52.2 96.2 29.4 8.0 70.6 92.0 .54 89.6 14.9

servations. For example, specific indications for testing may occur disproportionately in women with repeated NS Ts. In addition, the prevalence of nonreactive NS Ts would presumably be lower if repeated cases are included; those with a nonreactive NST may be delivered as a result of testing and thus never again contribute to the pool of NSTs studied. To explore the analytic impact of this issue, we studied all auscultated acceleration tests performed with repeated NSTs on the same women during the data collection period. For this analysis, 154 NSTs on the 100 previously described women were studied. The following variables were statistically significant in the logistic regression analysis (Table II): gestational age (p < 0.0001 ), test indication of previous pregnancy loss (p < 0.001 ), and the auscultated acceleration test (p < 0.001). The summary classification of these data is given in Table Ill. The classification table reads as follows: sensitivity, 52.2%; specificity, 96.2%; false-positive rate, 29.4%; and false-negative rate, 8.0%. The prevalence of a nonreactive NST (14.9%) was slightly lower compared with that of the previous analysis with initial NSTs only (16.0%). The kappa statistic for agreement was 0.54 in this analysis indicating agreement similar to that found in the earlier analyses based on the first 100 cases. Significant differences in sensitivity and false-positive results were observed between the two analyses. Additionally, the test indication of previous

1166

Paine, Johnson, and Alexander

l\m·cmhcr l 988

.\m

J Obs tel c;, nccol

Table IV. Summary classification of earlier work on auscultated acceleration test* and data from this study n = 74*

Sensitivity (%) Specificity (%) False-positive rate (%) False-negative rate (%) Positive predictive value (%) Negative predictive value (%) Kappa statistic Agreement(%) Prevalence of nonreactive NSTs (%) *From Paine et al.

75.0 92.8 62.5 1.5 37.5 98.5 .46

91.9 5.4

n =JOO

75.0 97.6 14.3 4.7 85.7 95.3 .54 89.6 16.0

7

pregnancy loss replaced race as a significant predictor of NST results. Comment

As in previously reported work on the accuracy of auscultation of FHR accelerations, 6 findings here suggest that accelerations of the FHR can be accurately auscultated. The sensitivity of the auscultated acceleration test when predicting the NST was similar to earlier studies using external stimulation to elicit fetal movement (sensitivity 75%).7 As seen in Table IV, using vibratory acoustic stimulation in this study yielded a decrease in the false-positive rate and a small improvement in specificity. Claims by others, 10 that fewer falsepositive results occur with the use of vibratory acoustic stimulation, were supported by this study. These results should encourage the use of vibratory acoustic stimulation over external stimulation during auscultated acceleration test. However, also evident in Table IV is the finding that the false-negative rate was increased with vibratory acoustic stimulation. This may be due to baseline changes or tachycardia. 13 Since a relatively high sensitivity with a minimum of false-negative results is desirable in any screening test for antepartum fetal well-being, further study is needed to explore other ways in which the validity of the auscultated acceleration test could be improved. Sensitivity may improve if auscultated acceleration test criteria were redefined as the auscultation of two accelerations in ten minutes or if NST criteria were reduced to one acceleration as an indicator of fetal health as proposed by Mendenhall. 14 Independence of observations is a basic underlying assumption of multiple regression statistical models. The inclusion of repeated NSTs from the same patient into the population of "cases" (e.g., NSTs) for analysis violates this assumption and, as observed in the results of this investigation, can result in disparate findings. We included data in Table III to show the potential errors such methodology can produce and encourage disuse of such methods in future analysis of fetal assessment techniques.

Previous studies of the auscultated acceleration test were performed on women between 35 and 43 weeks' gestation 6 7 and on women who had a significantly lower incidence of nonreactive NSTs (Table IV). Results from the analysis of these data indicated that testing at lower gestational ages results in a greater likelihood that the NST will be nonreactive. Considering these factors, as well as clinical questions regarding the appropriateness of applying the usual NST reactivity criteria in earlier gestational ages, 15 further assessment of the auscultated acceleration test should be undertaken analyzing its validity in specific gestational ages between 28 and 43 weeks. The same scrutiny should be applied to further evaluation of race in prediction of NST results because findings in this study indicated that race may play an important role. Studies with larger sample sizes are needed to evaluate the effect of gestational age, NST indications, auscultated acceleration test criteria for reactivity, and auscultated acceleration test testing time on the overall efficacy of the auscultated acceleration test in predicting NST results. However, the auscultated acceleration test continues to show potential as an initial screening test for fetal health and the auscultation of FHR accelerations represents a promising area of exploration as an alternative to the electronically monitored nonstress test, particularly for those women who have no access to electronic monitoring. We wish to acknowledge Deedra Rafkin, RN, BSN, Cathy Treanor, RNC, BSN, and Mildred Hairston of the Johns Hopkins Hospital Fetal Assessment Center for their assistance in data collection, and to Carol Gray, RN, BSN, of Towson State University for her assistance in data analysis. REFERENCES I. Evertson L, Gauthier RJ, Shifrin BS, Paul RH. Antepartum fetal heart rate testing. I. Evolution of the nonstress test. AM J OBSTET GYNECOL 1979; 133:29. 2. Devoe LD, Castillo RA, Sherline DM. The nonstress test as a diagnostic test: a critical reappraisal. AM J 0BSTET GYNECOL 1985;152:1047. 3. Devoe LD, Morrison], Martin], et al. A prospective com-

Volume 159 t\umber 5

4. 5. 6. 7. 8. 9.

Auscultated acceleration test with vibratory acoustic stimulation

parative study of the extended nonstress test and the nipple stimulation contraction stress test. AM J OBSTET GvNECOL 1987;157:531. Baskett T, Boyce C, Lohre M. Simplified antepartum fetal hear~ assessment. Br J Obstet Gynecol 1981;88:397. O'Leary J, Mendenhall H, Andrinopoulos G. Comparison of auditory versus electronic assessment of antenatal fetal welfare. Obstet Gynecol 1980;56:244. Paine LL, Payton RG, Johnson TRB. Auscultated FHR accelerations. Part I. Accuracy and documentation. J Nu rs Midwif 1986;31:68. Paine LL, Johnson TRB, Turner MH, Payton RG. Auscultated FHR accelerations. Part 2. An alternative to the nonsfress test. J Nurs Midwif 1986;31:73. Shy KK, Larson EB, Luthy DA. Evaluating a new technology: the effectiveness of electronic fetal heart rate monitoring. Am Rev Public Health 1987;8: 165. Serafini P, Lindsay MBJ, Nagey DA, Pupkin MJ, Tseng P, Crenshaw C. Antepartum fetal heart rate response to sound stimulation: the acoustic stimulation test. AM J OBSTET GYNECOL 1984; 148:41.

1167

10. Smith CV, Phelan JP, Platt LD, Broussard P, Paul RH. Fetal acoustic stimulation testing. II. A randomized clinical comparison with the nonstress test. AM J OBSTET GvNECOL 1986;155:131. 11. Fleiss JL. Statistical methods for rates and proportions. New York: John Wiley & Sons, 1981. 12. Thacker SB, Berkelman RL. Assessing the diagnostic accur.acy and efficacy of selected antepartum fetal surveillance techniques. Obstet Gynecol Surv 1986;41:121. 13. Gagnon R, Hunse C, Carmichael L, Fellows F, Patrick J. External vibratory acoustic stimulation near term: fetal heart rate variabiiity responses. AM J 0BSTET GYNECOL 1987;156:323. 14. Mendenhall H, O'Leary J, Phillips K. The nonstress test: the value of a single acceleration in evaluating the fetus· at risk. AM J OBSTET GYNECOL 1980;136:87. 15. Gagnon R, Campbell K, Patrick]. Patterns of human fetal heart rate accelerations from 26 weeks to term. AM J OBSTET GYNECOL 1987;157:743-8.