- Email: [email protected]
The effect of maternal glucose administration on the specificity of the nonstress test
The effect of maternal glucose administration on the specificity of the nonstress test BRYAN RICHARDSON, M.D.* MARY LOV BRIGGS. R.N. CHRIS TOOMEY, R.N. KATHY JOHNSON BURRY, R.N. JOHN PATRICK O'GRADY, M.D. Portland, Oregon To determine whether maternal glucose administration can decrease the incidence of false positive nonstress tests, 296 nonstress tests were performed on 235 high-risk obstetric patients in a prospective controlled study. Patients were alternately given a 50 gm oral glucose drink or an equal volume of water 30 minutes prior to the commencement of each test. Among "fed" patients (last meal within 2 hour of the nonstress test) whether receiving glucose or water, and "fasted" patients who received glucose, there was no significant difference in the percentage of reactive tests at either 20 minutes (65.2"k) or 40 minutes {87.3%) of testing. However, patients fasting and receiving water had a significantly decreased percentage of reactive tests, both at 20 minutes (48.3%, P < 0.01) and at 40 minutes (76.7%, P < 0.05). Glucose administered to fasted patients resulted in an increase in the incidence of reactive tests, although this was not statistically significant. Glucose administration had no effect on the nonstress test results when administered to fed patients. (AM. J. OesTET. GvNecoL 145:141, 1983.)
THE NONSTRESS TEST has become an important method for evaluating fetal health in high-risk obstetric patients. A reactive nonstress test with fetal movement and associated fetal heart rate acceleration has been shown to be an accurate indicator of a healthy fetus. 1 • 2 However, fetal movements have been demonstrated to occur in a cyclic rest-activity pattern that resembles the cyclic quiet and acti>·e sleep states of the healthy neonate. a· 4 A high incidence of nonreactive nonstress tests witb subsequent good fetal outcome, or false positive nonstress tests, has been attributed in part to the fetal rest-activity cycle. 2 It has been suggested that a nonreactive nons tress From the Obstetrical Special Studies Unit, Department of Ob,\tetrics and Gynecology, Oregon Health Sciences U nit•ersity. · Presented at the Thirty-eighth Annual Meeting of Thf Society of Obstetricians and Gynaecologists of Canada, Toronto, Ontario, CaMda,june 15-19, 1982.
Reprint requestr: Bryan S. Richardson, M.D., Department of Obstetrics and Gynecology, St. Joseph's Hospital, 268 Grost>e1wr, London, Ontario, Canada N6A 4V2. *Supported by a Canadian Medical Research Council F ellnwship.
0002-937RI83/020141+06$00.60/0
©
1983 The C. V. Mo~by Co.
test be repeated after a maternal meal, since the higher glucose levels may stimulate the healthy, but inactive fetus, but not affect the inactive, compromised fetus. 5 However, the effect of glucose on fetal activity is not clear. Two studies in which external tocodynamometrv and maternal perception of movement were used reported a significant increase in compound fetal movements subsequent to maternal glucose administration. 6 • 1 A third study in which real-time ultrasonic visualization of fetal activity was used reported similar findings. 8 However, two additional studies with real-time ultrasound noted only an increase in fetal breathing movements subsequent to maternal glucose administration, and no change in the incidence of gros~ fetal bodv moven1ents. 9 · 10 The purpose of this study was to determine, in a clinical setting, whether maternal glucose administration could improve the specificity of the nonstress test by decreasing the incidence of false positin~ nonstress tests.
Methods N onstress testing was performed by two specially trained nurses (M. L. B., C. T.) on selected high-risk 141
142 Richardson et al.
January 15. 1983 Am. J. Obstet. Gynecoi.
Table I. Test indications No.
Indications
Post dates Suspected intrauterine growth retardation Decreased fetal movement Hypertension Previous stillbirth Third-trimester bleeding Miscellaneous Total
No.
receiving glucose
recennng water
68 35
74 29
21.8
15
16
10.6
12
6.1 4.1
2
6 4 1
7
16
147
7.9
146
100.0
8
Percent 48.5
LO
patients referred to the Obstetrical Special Studies Unit ofthe Oregon Health Sciences University. Patients with diabetes mellitus were excluded. Indications for testing are listed in Table I. The gestational age of patients tested ranged from 28 to 42+ weeks. The time of the patient's last meal was noted. Patients then alternately received either a 50 gm oral glucose drink* or an equal volume of water. Nonstress testing commenced 30 minutes later. Informed consent was obtained in all cases. The research protocol was approved by the University Human Research Committee. Patients were placed in the semi-Fowler position, and the blood pressure was taken at 10-minute intervals to screen for supine hypotension. External fetal heart rate monitoring techniques were used with abdomina! wall fetal electrocardiograph or Doppler ultrasound.t Areactive nonstress test was defined as four or more accelerations in the fetal heart rate of~ 15 beats per minute lasting~ 15 seconds in association with fetal movement during a 20-minute interval. If the pattern was nonreactive in the first 20 minutes, monitoring was continued for a second 20-minute interval. Recordings were terminated when four fetal heart rate accelerations with fetal movement occurred in a 20-minute period or after 40 minutes. For the purpose of this study, the initial 20-minute period of recording and subsequent 20-minute period, if necessary, were compared between the two groups. Patients with nonreactive tests were considered for a repeat nonstress test the same day, oxytocin challenge test, or delivery, depending on the clinical situation. Reactive tests were repeated in I week if clinically indicated. Clinical information was collected prospectively, and analysis of recordings was performed by one of the *Dolce C Glucose Tolerance Test Beverage, Hopping Bottling Company, Sunnyvale, California. tHewlett-Packard 8030A, Hewlett Packard, Waltham, Massachusetts.
authors (B. R.) without clinical information or knowledge of the administration of glucose or water. Patients were managed by attending physicians who had access to nonstress test recordings but not to the interpretation of recordings presented in this study. Statistical analysis of data was by ·l analysis. Differences were considered to be significant when the P value was <0.05.
Results During the period from September, 1980, to April, 1982, 296 selected nonstress tests were performed on 235 patients. The details of testing and outcome for neonates with low Apgar scores (<7 at 5 minutes), fetal distress in labor necessitating cesarean section, and perinatal morbidity/mortality are listed in Table II. All other fetuses had good outcomes as defined by 5minute Apgar scores, cord arterial pH, and the absence of neonatal morbidity and mortality. Patients 1 and 2 were thought to have true positive nons tress tests indicative of fetal compromise at the time of testing and, thus, were removed from further statistical evaluation. Patients 3 through 8 were thought to have appropriately reactive nonstress tests indicative of good fetal condition at the time of study, although not necessarily predictive of the subsequent course during active labor. Patient 9 was thought to have a false positive nonstress test, since prior to delivery 7 days later, two reactive nonstress tests, not performed under the study protocol, intervened. Thus, for the purposes of this study, 293 nonstress tests performed on 233 patients were compared for the percentage of reactive and nonreactive nonstress tests between the group given glucose and the group given water. Among the 14 7 patients who received 50 gm of glucose orally, 63.3% of the nonstress tests were reactive within 20 minutes, whereas 86.5% were reactive with 40 minutes of testing (Table III). This result was not significantly different from that in the 146 control patients who received water, in whom 56.9% and 81.7% of the tests were reactive with 20 and 40 minutes of study, respectively (Table III). The patients were further subdivided into two groups: fed (last meal within 2 hours of nonstress test) and fasted (last meal more than 2 hours prior to nonstress test). For the patients who received glucose, no significant difference in the percentage of reactive tests was noted between the fed and fasted patients, either at 20 minutes or 40 minutes of testing (Table IV). However, for the control group receiving water, patients who had eaten within 2 hours had a significantly higher percentage of reactive tests with 20 minutes of testing: fed, 70.2%; fasted, 48.3% (P < 0.01) (Table V). With 40 minutes of testing, although a higher percentage of
Effect of maternal glucose on nonstress test
Volume 145 Number 2
143
Table II. Low Apgar score, fetal distress, perinatal morbidity/mortaiity Patient
2
N onstress test result; delivery interval (days)
Water-Nonreactive Water-Nonreactive 30 and 27 Water-Nonreactive :~
3
Glucose-Reactive 3
4
Water-Reactive I
5
Glucose-Reactive 34
6
Glucose-Reactive
7
Water-Reactive
I
Perinatal morbidity/mortality
Pregnancy complication; labor tracing
Chronic hypertension-propranolol 160 mg daily; fetal death at 36 wk
010
1,470
Fetal death
Cholestasis of pregnancy; late decelerations in early labor-cesarean section Decreased fetal movement; flat fetal heart rate baseline and failure to progress-cesarean section; cord pH 7.31 Suspected intrauterine growth retardation; late decelerations and failure to progress-cesarean section; cord pH 7.22 Suspected intrauterine growth retardation; late decelerationscesarean section; cord pH 7.22 Suspected intrauterine growth retardation; abruptio placentae at 34 weeks-cesarean section; cord pH
4/8
3,800
Neonatal death, tricuspid atresia
2/6
3,310
Meconium aspiration, {3-hemolvtic streptococcus pneumonia
5/9
2,520
Home on fifth day
3/8
2,570
Neonatal Intensive Care Unit, 12 days, mild necrotizing entero-
7/9
1,940
Mild respiratory distress syn· drome, necrotizing enterocolitis
4/6
2.730
Home on fifth day
9/9
2,840
Home on sixth day
5/9
1,600
Respiratory distress syndrome, persistent fetal circulation, neonatal death
colitis
7.33 10
8
Water-Reactive
9
Glucose-Nonreactive 7
60
Abdominal cerclage; amniocentesis at 36 wk bloody-spontaneous contraction stress test equivocal; repeat cesarean section Thalassemia; late decelerationscesarean section Eclampsia at 33 wk; cesarean section for maternal indications
Table III. Nonstress test results for group given glucose and group given water 40 min*
20 min
I
Reactive nonstress test
Glucose (N = 147) Water (;'If= 146)
Non reactive nonstress test
Reactive rwnstress teJt
54 63
122 (86.5%) 116 (81.7%)
93 (63.3%) 83 (56.9%)
I
Nonreactive nonstress test 19
26
NS
NS *Ten studies were of inadequate duration.
Table IV. Nonstress test results for fed and fasted groups administered glucose 40 min*
20 min Reactive nonstress test Fed, glucose
{I-'~
= 34)
Fasted, glucose (N
=
113)
.....
I
Nonreactive nonstress test
f"''t\ f)'.(Jf_ \ \IV•V/0)
10
69 (61.1 o/c)
44
NS
Reactive nonstress test
I
(84.9%) 94 (87.0%)
28
Nonreactive nonstres.v test 5 14
NS
*Six studies were of inadequate duration.
reactive tests was noted in the fed patients, this was not significant (Table V). When the glucose group was compared to the control group, with notation of the time of the last meal, no significant difference in the percentage of reactive tests
at 20 minutes was noted among fed patients whether receiving glucose or water and fasted patients who received glucose. Those fed patients and fasted patients who received glucose had a significantly higher percentage of reactive tests (65.2%) than fasted patients
144 Richardson et al.
January 15, 1983 Am. J. Obstet. Gynecol.
Table V. Nonstress test results for fed and fasted groups administered water 20 rnin
I
Reactive nonstress test
Fed, water (N = 57) Fasted, water (N = 89)
40 (70.2%) 43 (48.3%)
40 min* N lYYI.reactive nonstress test
Reactive nonstress test
17 46
50 (89.3%) 66 (76.7%)
x2 = 6.77
1
NlYYI.reactive nonstress test 6 20
NS
p < 0.01
*Four studies were of inadequate duration.
Table VI. Nonstress test results with redefined reactive criterion rate accelerations 15 bpm for~ 15 seconds/20-minute interval
~2
fetal movements with fetal heart
20min
Reactive nonstress test
Fed, glucose (N 34) Fed, water (N == 57) Fasted, glucose (N = 113) Fasted, water (N =89)
29 51 93 71
(85.3%) (89.5%) (82.3%) (79.8%)
I
40 min* NlYYI.reactive rumstress test 5 6 20 18
Reactive nonstress test 33 56 107 86
(97.1%) (98.3%) (96.4%) (96.6%)
I
N lYnreactive nonstress test l
1
4
3
*Two studies were of inadequate duration. who received water (48.3%) (P < 0.01). However, the percentage of reactive tests in only the fasted patients receiving glucose, although increased, was not significantly different from that of the fasted patients receiving water. With 40 minutes of testing, the fed patients and fasted patients receiving glucose had a significantly higher percentage of reactive tests ( 87.3%) than did the fasted patients who received water (76.7%) (P < 0.05). Again, the percentage of reactive tests in only the fasted patients who received glucose, although increased, was not significantly different from that of the fasted patients who received water. Recordings were further analyzed with the redefined criterion of two accelerations associated with fetal movement in a 20-minute period in order to be classified as a reactive test. The three nonreactive nonstress test recordings judged to be appropriately nonreactive and indicative of fetal compromise would still be scored as being nonreactive. With this criterion, the groups administered either glucose or water, whether fed or fasted, showed similar percentages of reactive tests after both 20 minutes and 40 minutes of testing (Table VI). Comment
This study in a high-risk obstetric population demonstrated a nonreactive rate for the nonstress test of 16.8% with 40 minutes of testing. Comparisons with the nonstress test literature are made difficult by the lack of uniformity among study populations and criteria for a reactive nonstress test. Evertson and asso-
ciates 11 reported a nonreactive rate of 34% with use of the criteria of five accelerations in a 20-minute period studied for up to 40 minutes, whereas Weingold and associates 12 reported a nonreactive rate of 5.8% with use of the same criteria. Using reactive criteria similar to those used in this study, Phelan2 reported a nonreactive rate of 14.0%. Of the 48 nonreactive tests, only three were judged to be true positives indicative of subsequent fetal or neonatal compromise which might be predicted at the time of the test. Therefore, of the other 45 nonreactive tests deemed to be false positives, only one manifested a subsequent poor outcome (Patient 9, as noted in Table II). The high proportion of nonreactive nonstress tests judged to be false positive may be attributed, in part, to the practice at our institution of using the oxytocin challenge test rather than the nonstress test to monitor diabetic patients, and severe intrauterine growth retardation and hypertensive pregnancies. This study demonstrated an increased percentage of nonreactive tests in those control patients who fasted for more than 2 hours prior to their study, as compared to those patients who had eaten within 2 hours of their study. Fifty grams of glucose administered orally to fasted patients 30 minutes prior to their study resulted in an increase in the incidence of reactive nonstress tests by 20 minutes, but only to a level intermediate between that of fed and fasted patients. This 30-minute interval may not be long enough for orally administered glucose to maximally increase the per-
Volume 145 Number 2
centage of reactive tests during 20 minutes of testing. When monitored over 40 minutes, fasted patients who were given glucose had a reactive rate similar to that of fed patients, and this group as a whole had a significantly higher percentage of reactive tests than did fasted patients who were given water. Glucose had no effect on the nonstress test results when administered to fed patients. The nonstress test as a measurement of fetal health depends upon the presence of fetal movement with associated fetal heart rate acceleration. Studies by Miller and associates 6 and Aladjem and associates 7 in fasted patients showed a significant increase in compound fetal movement for up to 3 hours after maternal glucose administration. In both of these studies, fetal movements were recorded by way of an external tocodynamometer placed on the maternal abdomen and identified bv maternal report. No attempt was made to control for observer error in interpretation of movement. Gelman and associates, 8 using real-time ultrasound, noted a similar increase in fetal movement I hour after maternal intravenous injection of 25 gm of glucose. However, observation periods were of short duration. and extreme variability is known to exist in fetal mowment, depending on the length of observation.14 In two additional controlled studies with prolonged real-time ultrasound observation, in which observers were blinded as to the glucose administration, Natale and associates 9 and Bocking and associates 10 noted only an increase in fetal breathing movements and no change in the incidence of gross fetal body movements. In both of those studies, only fetal rolling movements and fetal stretching movements of the trunk were recorded as gross fetal body movements. Isolated movements of fetal limbs were not identified. Timor-Tritsch and colleagues 13 showed that fetal body movements are strongly associated with fetal heart rate accelerations: with fetal movements of at least 1 second or more, 98% were associated with fetal heart rate acceleration of at least I 0 bpm. Although the present study did not measure fetal body movement, but rather fetal movement with associated fetal heart rate acceleration as required for a reactive nonstress test, the increase in reactive tests in fed patients or with the administration of glucose to fasted patients is suggestive of an effect on fetal body movement. It is possible, however, that glucose might affect the relationship to associated fetal heart rate acceleration rather than fetal movement. Studies of human fetal activity with maternal perception and tocodynamometers, and of fetal heart rate with maternal abdominal wall fetal electrocardiograph have demonstrated the existence of alternating fetal
Effect of maternal glucose on nonstress test
145
rest-activity periods that resemble the rest-activity sleep state patterns measured in newborn infants. 3 • 4 Campbell and associates 15 used time series analysis to demonstrate that episodes of gross fetal body movements normally last approximately 25 minutes and recur in intervals approximately every 60 minutes. The high incidence of false positive nonstress tests, which decreases the accuracy of identifying the compromised fetus, has been attributed, in part, to the fetal restactivity cycle. 2 In the human neonate, active sleep is significantly more prevalent immediately after a waking accompanied by feeding than after waking alone. 1' 1 Studies in the fetal Iamb with chronically implanted catheters and cortical electrodes have demonstrated a significant decrease in low-voltage electrocortical activity as seen in active sleep, with maternal fastingY Subsequent infusion of glucose increased the incidence of low-voltage electrocortical activity, but only to that level normally seen during the maternal fed state. If the fetal restactivity cycle represents changes in sleep state in the human fetus, it is possible that glucose might affect the incidence of reactive nonstress tests by increasing the time spent in fetal active states. It is also possible that the cyclic nature of fetal activity might he encrained after maternal meals or the administration of glucose, with the earlier onset of an active state rather than an increase in fetal activity. The incidence of false positive nonstress tests has been reported to be decreased by reducing the number of fetal heart rate accelerations with fetal movements required to define fetal reactivity during observation intervals. 11 In this study, with the redefined criteria of two movements with acceleration for a reactive test, sensitivity of the test was not sacrificed, since the three true positive nonstress tests would still be scored as nonreactive. The reactive rate for the fasted patients who were given water increased to 79.8% and 96.6% with 20 and 40 minutes of testing, respectively. This was not significantly different from the reactive rates for the fed patients and fasted patients who were given glucose, thus suggesting that, even with fasting, most fetuses in 40 minutes will have at least one 20-minute period with two movements and associated heart rate changes. The specificity of the nonstress test would appear to be more effectively improved with the redefined reactive criteria than by maternal glucose administration, although statements in regard to the sensitivity of the nonstress test are limited bv the small number of true positive tests. It is important to note that, although maternal fast· ing increased the incidence of false positive nonstress tests in this study, glucose administration was of no
146
Richardson et al.
more benefit in reducing this incidence than was simple testing of patients while in a fed state. Even with feeding or glucose administration and the redefined reactive criteria of two movements with acceleration, a small number of healthy fetuses will remain nonreac-
January 15, 1983 Am. J. Obstet. Gynecol.
tive with 40 minutes of testing. It is important to recognize the natural variability in the fetal rest-activity cycle and allow a more extended period of observation in these cases, as reported by Brown and Patrick, 1s before diagnosing the pathologically nonreactive fetus.
REFERENCES 1. Keegan, K. A., and Paul, R. H.: Antepartum fetal heart rate testing. IV. The nonstress test as a primary approach, AM.j. 0BSTET. GYNECOL. 136:75, 1980. 2. Phelan, J. P.: The nonstress test: A review of 3,000 tests, A,\.1, j. 0BSTET. GYNECOL. 139:7, 1981. 3. Junge, H. D.: Behavioral states and state related heart rate and motor activity patterns in the newborn infant and the fetus antepartum-A comparative study. I. Technique, illustration of recordings, and general results, J. Perinat. Med. 7:85, 1979. 4. Timor-Tritsch,J. E., Dierker, L.J., Hertz, R. H., Deagan, i\. C., and Rosen, M. G.: Studies of antepartum behavioral state in the human fetus at term, AM. J. OBSTET. GYNECOL. 132:524. 1978. 5. Keegan, K. A., Paul, R. H., Broussard, P.M., McCart, D., and Smith, M.A.: Antepartum fetal heart rate testing. V. The nonstress test-An outpatient approach, AM. J. 0BSTET. GYNECOL. 136:81, 1980. 6. Miller, F. C., Skiba, H., and Klapholz, H.: The effect of maternal blood sugar levels on fetal activity, Obstet. Gvnecol. 52:662. 1978. 7. Aladjem, S., Antonio, F., Rest,J., Gull, K., and O'Connor, M.: Effect of maternal glucose load on fetal activity, AM. j. 0BSTET. GYNECOL. 134:276, 1979. 8. Gelman, S. R., Spellacy, W. N., Wood, S., Birk, S. A., and Buhi, W. C.: Fetal movements and ultrasound: Effect of maternal intravenous glucose administration, AM. J. 0BSTET. GYNI'.COL. 137:459, 1980. 9. Natale, R .. Patrick ]., and Richardson, B.: Effects of human maternal venous plasma glucose concentrations on fetal breathing movement, AM.]. OssTET. GYNECOL. 132:36, 1978. 10. Bocking, A., Adamson, L., Cousin, A., Campbell, K.,
11. 12. 13.
14.
15.
16.
17.
18.
Carmichael, L., Natale R., and Patrick J .: Effects of intravenous glucose injections on human fetal breathing movements and gross fetal body movements at 38.40 weeks' gestational age, AM. J. OssTET. GYNECOL. 142: 606, 1982. Evenson, L. R., Gauthier, R. ]., Schifrin, B. S., and Paul, R. H.: Antepartum fetal heart rate testing. I. Evolution of the nonstress test, AM. J. OBSTET. GYNECOL. 133:29, 1979. Weingold, A. B., Yonekura, M. L., and O'Kieffe, J.: Nonstress testing, AM. J. 0BSTET. GYNECOL. 138:195, 1980. Timor-Tritsch, I. E., Dierker, L. j., Zador, I., Hertz, R. H., and Rosen, M.G.: Fetal movements associated with fetal heart rate accelerations and decelerations, AM. J. 0BSTET. GYNECOL. 131:276, 1978. Patrick J ., Campbell K., Carmichael L., Natale, R., and Richardson, B.: Patterns of gross fetal body movements over 24 hour observation intervals during the last 10 weeks of pregnancy, AM.]. OssTET. GYNECOL. 142:363, 1982. Campbell, K., MacNeill, I.. and Patrick].: Time series analysis of ultrasonic observations of gross fetal body movements during the last 10 weeks of pregnancv, Ultrasonic Imaging 3:330, 1981. Harper, R. M., Hoppenbrouwers, T., Bannett, D., Hodgman,]., Sterman, M. B., and McGinty, D.J.: Effects of feeding on state and cardiac regulation in the infant, Dev. Psychobiol. 10:507, 1977. Richardson, B .. Hohimer, R. A., Mueggler, P., and Bissonnette, ].: Effects of glucose concentration on fetal breathing movements and electrocortical activity in fetal lambs, AM.]. 0BSTET. Gv:-!ECOL. 142:678, 1982. Brown, R., and Patrick].: The nonstress test. How long is enough? AM.]. 0BSTET. GYKECOL. 141:646, 1981.
THE NONSTRESS TEST has become an important method for evaluating fetal health in high-risk obstetric patients. A reactive nonstress test with fetal movement and associated fetal heart rate acceleration has been shown to be an accurate indicator of a healthy fetus. 1 • 2 However, fetal movements have been demonstrated to occur in a cyclic rest-activity pattern that resembles the cyclic quiet and acti>·e sleep states of the healthy neonate. a· 4 A high incidence of nonreactive nonstress tests witb subsequent good fetal outcome, or false positive nonstress tests, has been attributed in part to the fetal rest-activity cycle. 2 It has been suggested that a nonreactive nons tress From the Obstetrical Special Studies Unit, Department of Ob,\tetrics and Gynecology, Oregon Health Sciences U nit•ersity. · Presented at the Thirty-eighth Annual Meeting of Thf Society of Obstetricians and Gynaecologists of Canada, Toronto, Ontario, CaMda,june 15-19, 1982.
Reprint requestr: Bryan S. Richardson, M.D., Department of Obstetrics and Gynecology, St. Joseph's Hospital, 268 Grost>e1wr, London, Ontario, Canada N6A 4V2. *Supported by a Canadian Medical Research Council F ellnwship.
0002-937RI83/020141+06$00.60/0
©
1983 The C. V. Mo~by Co.
test be repeated after a maternal meal, since the higher glucose levels may stimulate the healthy, but inactive fetus, but not affect the inactive, compromised fetus. 5 However, the effect of glucose on fetal activity is not clear. Two studies in which external tocodynamometrv and maternal perception of movement were used reported a significant increase in compound fetal movements subsequent to maternal glucose administration. 6 • 1 A third study in which real-time ultrasonic visualization of fetal activity was used reported similar findings. 8 However, two additional studies with real-time ultrasound noted only an increase in fetal breathing movements subsequent to maternal glucose administration, and no change in the incidence of gros~ fetal bodv moven1ents. 9 · 10 The purpose of this study was to determine, in a clinical setting, whether maternal glucose administration could improve the specificity of the nonstress test by decreasing the incidence of false positin~ nonstress tests.
Methods N onstress testing was performed by two specially trained nurses (M. L. B., C. T.) on selected high-risk 141
142 Richardson et al.
January 15. 1983 Am. J. Obstet. Gynecoi.
Table I. Test indications No.
Indications
Post dates Suspected intrauterine growth retardation Decreased fetal movement Hypertension Previous stillbirth Third-trimester bleeding Miscellaneous Total
No.
receiving glucose
recennng water
68 35
74 29
21.8
15
16
10.6
12
6.1 4.1
2
6 4 1
7
16
147
7.9
146
100.0
8
Percent 48.5
LO
patients referred to the Obstetrical Special Studies Unit ofthe Oregon Health Sciences University. Patients with diabetes mellitus were excluded. Indications for testing are listed in Table I. The gestational age of patients tested ranged from 28 to 42+ weeks. The time of the patient's last meal was noted. Patients then alternately received either a 50 gm oral glucose drink* or an equal volume of water. Nonstress testing commenced 30 minutes later. Informed consent was obtained in all cases. The research protocol was approved by the University Human Research Committee. Patients were placed in the semi-Fowler position, and the blood pressure was taken at 10-minute intervals to screen for supine hypotension. External fetal heart rate monitoring techniques were used with abdomina! wall fetal electrocardiograph or Doppler ultrasound.t Areactive nonstress test was defined as four or more accelerations in the fetal heart rate of~ 15 beats per minute lasting~ 15 seconds in association with fetal movement during a 20-minute interval. If the pattern was nonreactive in the first 20 minutes, monitoring was continued for a second 20-minute interval. Recordings were terminated when four fetal heart rate accelerations with fetal movement occurred in a 20-minute period or after 40 minutes. For the purpose of this study, the initial 20-minute period of recording and subsequent 20-minute period, if necessary, were compared between the two groups. Patients with nonreactive tests were considered for a repeat nonstress test the same day, oxytocin challenge test, or delivery, depending on the clinical situation. Reactive tests were repeated in I week if clinically indicated. Clinical information was collected prospectively, and analysis of recordings was performed by one of the *Dolce C Glucose Tolerance Test Beverage, Hopping Bottling Company, Sunnyvale, California. tHewlett-Packard 8030A, Hewlett Packard, Waltham, Massachusetts.
authors (B. R.) without clinical information or knowledge of the administration of glucose or water. Patients were managed by attending physicians who had access to nonstress test recordings but not to the interpretation of recordings presented in this study. Statistical analysis of data was by ·l analysis. Differences were considered to be significant when the P value was <0.05.
Results During the period from September, 1980, to April, 1982, 296 selected nonstress tests were performed on 235 patients. The details of testing and outcome for neonates with low Apgar scores (<7 at 5 minutes), fetal distress in labor necessitating cesarean section, and perinatal morbidity/mortality are listed in Table II. All other fetuses had good outcomes as defined by 5minute Apgar scores, cord arterial pH, and the absence of neonatal morbidity and mortality. Patients 1 and 2 were thought to have true positive nons tress tests indicative of fetal compromise at the time of testing and, thus, were removed from further statistical evaluation. Patients 3 through 8 were thought to have appropriately reactive nonstress tests indicative of good fetal condition at the time of study, although not necessarily predictive of the subsequent course during active labor. Patient 9 was thought to have a false positive nonstress test, since prior to delivery 7 days later, two reactive nonstress tests, not performed under the study protocol, intervened. Thus, for the purposes of this study, 293 nonstress tests performed on 233 patients were compared for the percentage of reactive and nonreactive nonstress tests between the group given glucose and the group given water. Among the 14 7 patients who received 50 gm of glucose orally, 63.3% of the nonstress tests were reactive within 20 minutes, whereas 86.5% were reactive with 40 minutes of testing (Table III). This result was not significantly different from that in the 146 control patients who received water, in whom 56.9% and 81.7% of the tests were reactive with 20 and 40 minutes of study, respectively (Table III). The patients were further subdivided into two groups: fed (last meal within 2 hours of nonstress test) and fasted (last meal more than 2 hours prior to nonstress test). For the patients who received glucose, no significant difference in the percentage of reactive tests was noted between the fed and fasted patients, either at 20 minutes or 40 minutes of testing (Table IV). However, for the control group receiving water, patients who had eaten within 2 hours had a significantly higher percentage of reactive tests with 20 minutes of testing: fed, 70.2%; fasted, 48.3% (P < 0.01) (Table V). With 40 minutes of testing, although a higher percentage of
Effect of maternal glucose on nonstress test
Volume 145 Number 2
143
Table II. Low Apgar score, fetal distress, perinatal morbidity/mortaiity Patient
2
N onstress test result; delivery interval (days)
Water-Nonreactive Water-Nonreactive 30 and 27 Water-Nonreactive :~
3
Glucose-Reactive 3
4
Water-Reactive I
5
Glucose-Reactive 34
6
Glucose-Reactive
7
Water-Reactive
I
Perinatal morbidity/mortality
Pregnancy complication; labor tracing
Chronic hypertension-propranolol 160 mg daily; fetal death at 36 wk
010
1,470
Fetal death
Cholestasis of pregnancy; late decelerations in early labor-cesarean section Decreased fetal movement; flat fetal heart rate baseline and failure to progress-cesarean section; cord pH 7.31 Suspected intrauterine growth retardation; late decelerations and failure to progress-cesarean section; cord pH 7.22 Suspected intrauterine growth retardation; late decelerationscesarean section; cord pH 7.22 Suspected intrauterine growth retardation; abruptio placentae at 34 weeks-cesarean section; cord pH
4/8
3,800
Neonatal death, tricuspid atresia
2/6
3,310
Meconium aspiration, {3-hemolvtic streptococcus pneumonia
5/9
2,520
Home on fifth day
3/8
2,570
Neonatal Intensive Care Unit, 12 days, mild necrotizing entero-
7/9
1,940
Mild respiratory distress syn· drome, necrotizing enterocolitis
4/6
2.730
Home on fifth day
9/9
2,840
Home on sixth day
5/9
1,600
Respiratory distress syndrome, persistent fetal circulation, neonatal death
colitis
7.33 10
8
Water-Reactive
9
Glucose-Nonreactive 7
60
Abdominal cerclage; amniocentesis at 36 wk bloody-spontaneous contraction stress test equivocal; repeat cesarean section Thalassemia; late decelerationscesarean section Eclampsia at 33 wk; cesarean section for maternal indications
Table III. Nonstress test results for group given glucose and group given water 40 min*
20 min
I
Reactive nonstress test
Glucose (N = 147) Water (;'If= 146)
Non reactive nonstress test
Reactive rwnstress teJt
54 63
122 (86.5%) 116 (81.7%)
93 (63.3%) 83 (56.9%)
I
Nonreactive nonstress test 19
26
NS
NS *Ten studies were of inadequate duration.
Table IV. Nonstress test results for fed and fasted groups administered glucose 40 min*
20 min Reactive nonstress test Fed, glucose
{I-'~
= 34)
Fasted, glucose (N
=
113)
.....
I
Nonreactive nonstress test
f"''t\ f)'.(Jf_ \ \IV•V/0)
10
69 (61.1 o/c)
44
NS
Reactive nonstress test
I
(84.9%) 94 (87.0%)
28
Nonreactive nonstres.v test 5 14
NS
*Six studies were of inadequate duration.
reactive tests was noted in the fed patients, this was not significant (Table V). When the glucose group was compared to the control group, with notation of the time of the last meal, no significant difference in the percentage of reactive tests
at 20 minutes was noted among fed patients whether receiving glucose or water and fasted patients who received glucose. Those fed patients and fasted patients who received glucose had a significantly higher percentage of reactive tests (65.2%) than fasted patients
144 Richardson et al.
January 15, 1983 Am. J. Obstet. Gynecol.
Table V. Nonstress test results for fed and fasted groups administered water 20 rnin
I
Reactive nonstress test
Fed, water (N = 57) Fasted, water (N = 89)
40 (70.2%) 43 (48.3%)
40 min* N lYYI.reactive nonstress test
Reactive nonstress test
17 46
50 (89.3%) 66 (76.7%)
x2 = 6.77
1
NlYYI.reactive nonstress test 6 20
NS
p < 0.01
*Four studies were of inadequate duration.
Table VI. Nonstress test results with redefined reactive criterion rate accelerations 15 bpm for~ 15 seconds/20-minute interval
~2
fetal movements with fetal heart
20min
Reactive nonstress test
Fed, glucose (N 34) Fed, water (N == 57) Fasted, glucose (N = 113) Fasted, water (N =89)
29 51 93 71
(85.3%) (89.5%) (82.3%) (79.8%)
I
40 min* NlYYI.reactive rumstress test 5 6 20 18
Reactive nonstress test 33 56 107 86
(97.1%) (98.3%) (96.4%) (96.6%)
I
N lYnreactive nonstress test l
1
4
3
*Two studies were of inadequate duration. who received water (48.3%) (P < 0.01). However, the percentage of reactive tests in only the fasted patients receiving glucose, although increased, was not significantly different from that of the fasted patients receiving water. With 40 minutes of testing, the fed patients and fasted patients receiving glucose had a significantly higher percentage of reactive tests ( 87.3%) than did the fasted patients who received water (76.7%) (P < 0.05). Again, the percentage of reactive tests in only the fasted patients who received glucose, although increased, was not significantly different from that of the fasted patients who received water. Recordings were further analyzed with the redefined criterion of two accelerations associated with fetal movement in a 20-minute period in order to be classified as a reactive test. The three nonreactive nonstress test recordings judged to be appropriately nonreactive and indicative of fetal compromise would still be scored as being nonreactive. With this criterion, the groups administered either glucose or water, whether fed or fasted, showed similar percentages of reactive tests after both 20 minutes and 40 minutes of testing (Table VI). Comment
This study in a high-risk obstetric population demonstrated a nonreactive rate for the nonstress test of 16.8% with 40 minutes of testing. Comparisons with the nonstress test literature are made difficult by the lack of uniformity among study populations and criteria for a reactive nonstress test. Evertson and asso-
ciates 11 reported a nonreactive rate of 34% with use of the criteria of five accelerations in a 20-minute period studied for up to 40 minutes, whereas Weingold and associates 12 reported a nonreactive rate of 5.8% with use of the same criteria. Using reactive criteria similar to those used in this study, Phelan2 reported a nonreactive rate of 14.0%. Of the 48 nonreactive tests, only three were judged to be true positives indicative of subsequent fetal or neonatal compromise which might be predicted at the time of the test. Therefore, of the other 45 nonreactive tests deemed to be false positives, only one manifested a subsequent poor outcome (Patient 9, as noted in Table II). The high proportion of nonreactive nonstress tests judged to be false positive may be attributed, in part, to the practice at our institution of using the oxytocin challenge test rather than the nonstress test to monitor diabetic patients, and severe intrauterine growth retardation and hypertensive pregnancies. This study demonstrated an increased percentage of nonreactive tests in those control patients who fasted for more than 2 hours prior to their study, as compared to those patients who had eaten within 2 hours of their study. Fifty grams of glucose administered orally to fasted patients 30 minutes prior to their study resulted in an increase in the incidence of reactive nonstress tests by 20 minutes, but only to a level intermediate between that of fed and fasted patients. This 30-minute interval may not be long enough for orally administered glucose to maximally increase the per-
Volume 145 Number 2
centage of reactive tests during 20 minutes of testing. When monitored over 40 minutes, fasted patients who were given glucose had a reactive rate similar to that of fed patients, and this group as a whole had a significantly higher percentage of reactive tests than did fasted patients who were given water. Glucose had no effect on the nonstress test results when administered to fed patients. The nonstress test as a measurement of fetal health depends upon the presence of fetal movement with associated fetal heart rate acceleration. Studies by Miller and associates 6 and Aladjem and associates 7 in fasted patients showed a significant increase in compound fetal movement for up to 3 hours after maternal glucose administration. In both of these studies, fetal movements were recorded by way of an external tocodynamometer placed on the maternal abdomen and identified bv maternal report. No attempt was made to control for observer error in interpretation of movement. Gelman and associates, 8 using real-time ultrasound, noted a similar increase in fetal movement I hour after maternal intravenous injection of 25 gm of glucose. However, observation periods were of short duration. and extreme variability is known to exist in fetal mowment, depending on the length of observation.14 In two additional controlled studies with prolonged real-time ultrasound observation, in which observers were blinded as to the glucose administration, Natale and associates 9 and Bocking and associates 10 noted only an increase in fetal breathing movements and no change in the incidence of gross fetal body movements. In both of those studies, only fetal rolling movements and fetal stretching movements of the trunk were recorded as gross fetal body movements. Isolated movements of fetal limbs were not identified. Timor-Tritsch and colleagues 13 showed that fetal body movements are strongly associated with fetal heart rate accelerations: with fetal movements of at least 1 second or more, 98% were associated with fetal heart rate acceleration of at least I 0 bpm. Although the present study did not measure fetal body movement, but rather fetal movement with associated fetal heart rate acceleration as required for a reactive nonstress test, the increase in reactive tests in fed patients or with the administration of glucose to fasted patients is suggestive of an effect on fetal body movement. It is possible, however, that glucose might affect the relationship to associated fetal heart rate acceleration rather than fetal movement. Studies of human fetal activity with maternal perception and tocodynamometers, and of fetal heart rate with maternal abdominal wall fetal electrocardiograph have demonstrated the existence of alternating fetal
Effect of maternal glucose on nonstress test
145
rest-activity periods that resemble the rest-activity sleep state patterns measured in newborn infants. 3 • 4 Campbell and associates 15 used time series analysis to demonstrate that episodes of gross fetal body movements normally last approximately 25 minutes and recur in intervals approximately every 60 minutes. The high incidence of false positive nonstress tests, which decreases the accuracy of identifying the compromised fetus, has been attributed, in part, to the fetal restactivity cycle. 2 In the human neonate, active sleep is significantly more prevalent immediately after a waking accompanied by feeding than after waking alone. 1' 1 Studies in the fetal Iamb with chronically implanted catheters and cortical electrodes have demonstrated a significant decrease in low-voltage electrocortical activity as seen in active sleep, with maternal fastingY Subsequent infusion of glucose increased the incidence of low-voltage electrocortical activity, but only to that level normally seen during the maternal fed state. If the fetal restactivity cycle represents changes in sleep state in the human fetus, it is possible that glucose might affect the incidence of reactive nonstress tests by increasing the time spent in fetal active states. It is also possible that the cyclic nature of fetal activity might he encrained after maternal meals or the administration of glucose, with the earlier onset of an active state rather than an increase in fetal activity. The incidence of false positive nonstress tests has been reported to be decreased by reducing the number of fetal heart rate accelerations with fetal movements required to define fetal reactivity during observation intervals. 11 In this study, with the redefined criteria of two movements with acceleration for a reactive test, sensitivity of the test was not sacrificed, since the three true positive nonstress tests would still be scored as nonreactive. The reactive rate for the fasted patients who were given water increased to 79.8% and 96.6% with 20 and 40 minutes of testing, respectively. This was not significantly different from the reactive rates for the fed patients and fasted patients who were given glucose, thus suggesting that, even with fasting, most fetuses in 40 minutes will have at least one 20-minute period with two movements and associated heart rate changes. The specificity of the nonstress test would appear to be more effectively improved with the redefined reactive criteria than by maternal glucose administration, although statements in regard to the sensitivity of the nonstress test are limited bv the small number of true positive tests. It is important to note that, although maternal fast· ing increased the incidence of false positive nonstress tests in this study, glucose administration was of no
146
Richardson et al.
more benefit in reducing this incidence than was simple testing of patients while in a fed state. Even with feeding or glucose administration and the redefined reactive criteria of two movements with acceleration, a small number of healthy fetuses will remain nonreac-
January 15, 1983 Am. J. Obstet. Gynecol.
tive with 40 minutes of testing. It is important to recognize the natural variability in the fetal rest-activity cycle and allow a more extended period of observation in these cases, as reported by Brown and Patrick, 1s before diagnosing the pathologically nonreactive fetus.
REFERENCES 1. Keegan, K. A., and Paul, R. H.: Antepartum fetal heart rate testing. IV. The nonstress test as a primary approach, AM.j. 0BSTET. GYNECOL. 136:75, 1980. 2. Phelan, J. P.: The nonstress test: A review of 3,000 tests, A,\.1, j. 0BSTET. GYNECOL. 139:7, 1981. 3. Junge, H. D.: Behavioral states and state related heart rate and motor activity patterns in the newborn infant and the fetus antepartum-A comparative study. I. Technique, illustration of recordings, and general results, J. Perinat. Med. 7:85, 1979. 4. Timor-Tritsch,J. E., Dierker, L.J., Hertz, R. H., Deagan, i\. C., and Rosen, M. G.: Studies of antepartum behavioral state in the human fetus at term, AM. J. OBSTET. GYNECOL. 132:524. 1978. 5. Keegan, K. A., Paul, R. H., Broussard, P.M., McCart, D., and Smith, M.A.: Antepartum fetal heart rate testing. V. The nonstress test-An outpatient approach, AM. J. 0BSTET. GYNECOL. 136:81, 1980. 6. Miller, F. C., Skiba, H., and Klapholz, H.: The effect of maternal blood sugar levels on fetal activity, Obstet. Gvnecol. 52:662. 1978. 7. Aladjem, S., Antonio, F., Rest,J., Gull, K., and O'Connor, M.: Effect of maternal glucose load on fetal activity, AM. j. 0BSTET. GYNECOL. 134:276, 1979. 8. Gelman, S. R., Spellacy, W. N., Wood, S., Birk, S. A., and Buhi, W. C.: Fetal movements and ultrasound: Effect of maternal intravenous glucose administration, AM. J. 0BSTET. GYNI'.COL. 137:459, 1980. 9. Natale, R .. Patrick ]., and Richardson, B.: Effects of human maternal venous plasma glucose concentrations on fetal breathing movement, AM.]. OssTET. GYNECOL. 132:36, 1978. 10. Bocking, A., Adamson, L., Cousin, A., Campbell, K.,
11. 12. 13.
14.
15.
16.
17.
18.
Carmichael, L., Natale R., and Patrick J .: Effects of intravenous glucose injections on human fetal breathing movements and gross fetal body movements at 38.40 weeks' gestational age, AM. J. OssTET. GYNECOL. 142: 606, 1982. Evenson, L. R., Gauthier, R. ]., Schifrin, B. S., and Paul, R. H.: Antepartum fetal heart rate testing. I. Evolution of the nonstress test, AM. J. OBSTET. GYNECOL. 133:29, 1979. Weingold, A. B., Yonekura, M. L., and O'Kieffe, J.: Nonstress testing, AM. J. 0BSTET. GYNECOL. 138:195, 1980. Timor-Tritsch, I. E., Dierker, L. j., Zador, I., Hertz, R. H., and Rosen, M.G.: Fetal movements associated with fetal heart rate accelerations and decelerations, AM. J. 0BSTET. GYNECOL. 131:276, 1978. Patrick J ., Campbell K., Carmichael L., Natale, R., and Richardson, B.: Patterns of gross fetal body movements over 24 hour observation intervals during the last 10 weeks of pregnancy, AM.]. OssTET. GYNECOL. 142:363, 1982. Campbell, K., MacNeill, I.. and Patrick].: Time series analysis of ultrasonic observations of gross fetal body movements during the last 10 weeks of pregnancv, Ultrasonic Imaging 3:330, 1981. Harper, R. M., Hoppenbrouwers, T., Bannett, D., Hodgman,]., Sterman, M. B., and McGinty, D.J.: Effects of feeding on state and cardiac regulation in the infant, Dev. Psychobiol. 10:507, 1977. Richardson, B .. Hohimer, R. A., Mueggler, P., and Bissonnette, ].: Effects of glucose concentration on fetal breathing movements and electrocortical activity in fetal lambs, AM.]. 0BSTET. Gv:-!ECOL. 142:678, 1982. Brown, R., and Patrick].: The nonstress test. How long is enough? AM.]. 0BSTET. GYKECOL. 141:646, 1981.