- Email: [email protected]
Patterns of human fetal heart rate accelerations from 26 weeks to term
Patterns of human fetal heart rate accelerations from 26 weeks to term Robert Gagnon, M.D.,* Karen Campbell, Ph.D., Cora Hunse, R.N., and JohJI Patrick, M.D.** London, Ontario, Canada Computerized analysis of the distribution of 2598 fetal heart rate accelerations in 83 healthy fetuses at 26 to 40 weeks' gestation demonstrated that the currently used definition of an acceleration as ;;.15 beats/min for ;;.15 seconds is applicable only after 30 weeks' gestational age in fetuses with a basal fetal heart rate of ,;128 beats/min. A significant negative correlation was found between the mean hourly basal fetal heart rate and the mean amplitude of fetal heart rate accelerations from 30 weeks to term. There was also a significant maturational process in the pattern of fetal heart rate and fetal heart rate accelerations ttlat occurred between 26 and 28 and between 30 and 32 weeks; this was characterized by a decrease in basal fetal heart rate, an increase in the amplitude of fetal heart rate accelerations, and an increase in long-term fetal heart rate variability. (AM J OasrET GYNECOL 1987;157:743-8.)
Key words: Nonstress test; fetal heart rate accelerations during third trimester; effect of gestational age on nonstress test Antepartum fetal heart rate (FHR) testing, including nonstress and contraction stress testing, has been se riously criticized 1• 2 because of a low positive predictive value, absence of strict and uniform standards for in terpretation, and the potential effect of gestational age on FHR reactivity. In addition, until recently accurate recording aQd interpretation of human FHR with Doppler ultrasound have presented m
Patients and methods Patients. informed consent was obtained from 89 healthy women with pregnancies between 26 and 40 weeks' gestational age. Eighty-three women and fetuses
From the Departments of Obstetrics and Gynecology, Physiology, and Epidemiology, University of Western Ontario, St. joseph's Hospital Research Institute. Supported by the Physicians' Services Incorporated Foundation and the Canadian Medical Research Council. Presented at the Seventh Annual Meeting of The Society of Perinatal Obstetricians, Lake Buena Vista, Florida, February 5-7, 1987. *Supported by a Fellowship from the Variety Club of Ontario. **Investigator in the Canadian Medical Research Council Group in Reproductive Biology. Reprint requests: Robert Gagnon, M.D., Department of Obstetrics and Gynecology, St. Joseph's Hospital, 268 Grosvenor St., London, Ontario N6A 4V2, Canada.
remained in good health, and six were excluded from the study. Five women were delivered of their infants before 37 weeks: Two had abruptio placentae, two had premature labor, and one was delivered of a stillborn infant at 35 weeks (4 weeks after study): All compli cations were of unknown cause. Another patient was lost to follow-up. The remaining 83 women had normal deliveries at term that confirmed gestational age and the good health of fetuses studied. The average ges tational age at delivery was 39.3 ± 0.1 weeks (SEM). The average weight of fetuses was 3350 ± 42 gm. The average 5-minute Apgar score was 9.3 ± 0.1 (mode 9; range 7 to 10). No Apgar score was less than 7 at 5 minutes. Umbilical arterial cord blood samples were obtained at birth in 58 of the 83 fetuses. The mean umbilical arterial Po 2 was 17.8 ± 1.0 mm Hg, Pco 2 was 50.0 ± 1.3 mm Hg, and pH was 7.27 ± 0.01. Ten (13.7%) women were delivered of their infants by ce sarean section: Two developed acute fetal distress in labor (5-minute Apgar scores of 8 and 10, respectively), one had cephalopelvic disproportion, one had a trans verse lie at the onset of labor, and six had repeat ce sarean sections. Eighty-one fetuses were in vertex pre sentation and two were breech at the time of delivery. Experimental design. The women received a stan dardized 800 kcal breakfast or lunch at 0800 or 1200 hours. All studies began 1 hour later and were con ducted in a quiet room with patients either sitting up right in bed or resting in a lateral recumbent position. Women were asked not to smoke for at least 4 hours before studies. Smoking or food intake by patients or observers was not permitted during the study. Women were studied continuously for 1 hour. Nine women participated in more than one study, and these were
743
744
CJ)
c:
Gagnon et al.
September 1987 Am J Obstet Gynecol
100
~
e CL>
Q;
36-40 wks • 33-35 wks • 30-32 wks e
80
(.) (.)
0
a:::
60
0
40
26-28 wks
•
I
.......
]
0 CL> 0>
c
20
0
CL>
(.)
' CL> Cl.
0 5
15
25
35
45
55
65
Amplitude of FHR accelerations (bpm) Fig. I. A cumulative histogram of 2598 FHR accelerations of ;a. 15 seconds' duration was plotted in successive increments of 5 beats/min to define the distribution of acceleration am plitudes in fetuses from 26 to 40 weeks' gestational age.
conducted in different gestational age groups. Twenty three studies were done at 26 to 28 weeks, 24 at 30 to 32 weeks, 26 at 33 to 35 weeks, and 28 at 36 to 40 weeks, for a total observation time of 101 hours. Fetal heart rate measurements. Fetal heart rate was recorded with a Hewlett-Packard 8040A (Hewlett Packard GMBH, Boeblinger, Federal Republic of Ger many) external monitor by use of a compact transducer and was analyzed on-line by a Sage II microprocessor (Sage Computer Technology, Reno, Nevada) as pre viously described.' The FHR intervals were examined by means of an artifact detection algorithm," and valid pulse intervals were averaged over 3.75 seconds. Dur ing on-line analysis a baseline was fitted to the heart rate record/ and accelerations of ;;;.J 0 beats/min for ;;;.}5 seconds above the baseline were recognized au tomatically. For each 10-minute interval the basal FHR (beats per minute) , the mean minute range (milliseconds),' and the number of FHR accelerations were calculated. The mean minute range value was calculated as the differ ence between the longest and the shortest RR intervals eyery minute and averaged for the length of observa tion. The mean minute range is a computer-derived index of long-term FHR variability.' Fetal breathing movements and gross fetal body movements were measured with a real-time ADR ul trasound scanner (Model 2130; Advanced Diagnostic Research Corp., Tempe, Arizona) and will be reported elsewhere. Data analysis. The basal FHR, the number of FHR accelerations of ;;;.JO beats/min for ;;;.15 seconds above the baseline, and amplitude of individual FHR ac celerations were determined on line in successive 10-minute intervals. The number of FHR acceler ations of ;;;.JS seconds' duration were also auto matically determined off-line for acceleration ampli-
tudes of 5 beats/min and in successive increments of 5 beats/min to a maximum amplitude of 60 beats/min. Portions of records that could not be analyzed owing to FHR signal loss of more than 30% or patient inter ruption were not included and made up 5.2% of total study time. The average signal loss of the remaining 96 hours included in this analysis was 6.9% ± 0.9%, 5.3% ± 0.6%, 5.7% ± 0.8%, and 4 .5% ± 0.5% at 26 to 28, 30 to 32, 33 to 35, and 36 to 40 weeks', respec tively. Results are presented as grouped means and stan dard errors of the mean. The significance of differ eqces in the means was examined by Bonferroni t test statistic tables to allow for multiple comparisons. 8 Stan dard regression analyses with the Pearson product moment correlations (r) were determined. Regression lines were compared for co-incidence by means of regression procedures described by Kleinbaum and Kupper! This analysis was performed with the package MINITAB (MINITAB Data Analysis Software, State College, Pennsylvania). Results Distribution of FHR accelerations. Fig. 1 is a cu mulative histogram plot of 2598 FHR accelerations of ;;;.JS seconds above the baseline in 83 fetuses between 26 and 40 weeks' gestational age. The percentage of total FHR accelerations was calculated in successive in crements of 5 beats/min to define the distribution of acceleration amplitudes. Of FHR accelerations between 30 and 40 weeks' gestation, 31.7% were ;;;.J5 beats/min for ;;;.}5 seconds. However, only 6.7% of FHR accel erations were ;;;.JS beats/min for ;;;.JS seconds at 26 to 28 weeks. The percentage of time intervals during which fewer than two FHR accelerations of ;;;.J5 beats/min for ;;;.15 seconds occurred is plotted in 10-minute increments (Fig. 2, A). From 30 to 40 weeks, 57.5 % of tracings had fewer than two FHR accelerations at 10 minutes and 2.7% at 60 minutes. However, at 26 to 28 weeks, 95.3% of recordings had fewer than two FHR accelerations at 10 minutes, 67.0% at 40 minutes, and 65.0% at 60 minutes, suggesting that most of the FHR accelerations in this group were <15 beats/min and cannot be de tected with an increase in the observation time. How ever, using a minimum amplitude of 10 beats/min for ;;;.}5 seconds to define an acceleration at 26 to 28 weeks, the length of time from the beginning of recordings until two FHR accelerations occurred was similar to that at 30 to 40 weeks (Fig. 2, A). In Fig. 2, B , it can be seen that with a decrease in the minimum amplitude to define an acceleration, there was a concomitant decrease in the percentage of 10-minute intervals during which fewer than two FHR accelerations occurred. This was observed at all ges tations from 26 to 40 weeks.
745
Patterns of FHR accelerations
Volume 157 Number 3
100
36-..0 wks
•
lQ-32 wks
• •
26-28 wks
80
150
e
ea::
33-35 .....
a:
i
140
.8
60
~
135
~
40
145
A
c:
j
A
20
130 20
0
10
0 -+-' 0
20
30
40
60
50
Time (minutes)
-+-'
Q)
0
c Q) u ..._
Q)
Q_
15
"'"
B
C:a:
2a:
100
:f .
O"l -+-'
.,, ,.,i
" u g .l!
'+
0
~~
~~ } .~
10
80
60
60
a:
55
.d E E
50
~~
45
J~
40
I~
c:c
2
B
20
::1
c
>
40 35
0
5
15
25
35
45
55
Amplitude of FHR accelerations (bpm)
Fig. 2. A, The percentage of time intervals during which fewer than two FH R accelerations of ;;.]5 beats/ min for ;;.]5 seconds occurred was plotted for all fetuses from 26 weeks to term. The percentage of time intervals during which fewer than two FHR accelerations of;;;.} 0 beats/min for ;;.]5 seconds was also plotted for fetuses from 26 to 28 weeks on the same time scale ( - - - - ). B, Cumulative percentage of I0-minute intervals dur ing which fewer than two FHR accelerations occurred was plotted in successive increments of 5 beats/min to define the minimum amplitude for an acceleration in all fetuses from 26 to 40 weeks.
Relationship between gestational age, basal FHR, and amplitude of FHR accelerations. Mean hourly ba sal FHR was 144.4 ± 1.3 beats/min at 26 to 28 weeks, which was significantly higher than at 30 to 32, 33 to 35, and 36 to 40 weeks (p < 0.05, p < 0.01, p < 0.01, respectively) (Fig. 3, A). Figure 3, B, demonstrated that there was a significant increase in the mean amplitude of FliR accelerations from 14.1 ± 0.5 beats/min at 26 to 28 weeks to 17.7 ± 0.8 beats/min at 30 to 32 weeks (p < 0.01), and this remained higher at 33 to 35 and 36 to 40 weeks (both p < 0.01) (Fig. 3, B). Finally, the 10-minute mean minute range was lower at 26 to 28 weeks than at 30 to 32, 33 to 3fi, and 36 to 40 weeks (all p < 0.01) (Fig. 3, C). It was important that there was no difference in the mean FHR, acceleration arne plitude, and mean minute range between 30 and 40 weeks' gestational age. There was a significant negative correlation between
26-28
3Q-32
33-35
36-40
Gestational age (weeks)
Fig. 3. A, The mean hourly basal FHR ( ± SEM) was plotted for fetuses at 26 to 28, 30 to 32, 33 to 35, and 36 to 40 weeks' gestation. B, The mean hourly amplitude of FHR accelera tions and, C, the mean 10-minute mean minute range (MMR) were also plotted on the same scale.
the mean hourly amplitude of FHR accelerations of ;;;.IO beats/min for ;:;;.15 seconds and the mean hourly basal FHR at 30 to 32, 33 to 35, and 36 to 40 weeks' gestation (Fig. 4, B to D), which demonstrated that within individual fetuses at 30 to 40 weeks the higher the mean hourly heart rate, the lower the height of individual FHR acceleration. No such correlation was found at 26 to 28 weeks (Fig. 4, A) . A comparison of the regression lines of the relation ship of FHR acceleration amplitudes and basal FHR at 30 to 40 weeks, by means of regression procedures described by Kleinbaum and Kupper, 9 confirmed that the regressions of Fig. 4, B to D, were coincident (slopes and intercepts were equal). Therefore these three sets of data were pooled, and Fig. 5 demonstrated the regression of FliR acceler
746 Gagnon et at.
September 1987 Am J Obstet Gynecol
35
2&-28 wks
A
30
30
25
25
20
20
... .., ..
15
8
35
30-32 wks r = -0.55 p < 0.01
.. .. . . . .· ..
15 ~
10
120
110
140
130
c
35
150
33-35 wks = -0.56 < 0.01
35
r p
30
110
160
25
25
20
20
.,.·.
15
. ....
120
130
150
160
36-40 wks = -0.66
D
r p
~
15
10
140
<
o.ooi
... . . ..·: ..
10 110
120
130
140
150
160
110
120
130
HO
150
160
Mean hourly basal fetal heart rate (bpm) Fig. 4. There was a significant negative correlation between the mean hourly amplitude of FHR accelerations and the mean hourly basal heart rate at all gestations from 30 to 40 weeks (B, C, D). No correlation was found at 26 to 28 weeks' gestation (A).
97.5 % of FHR accelerations with a basal FHR of ~ 143 beats/min in fetuses of 30 to 40 weeks' gestational age. Comment
Analysis of the distributior) of FHR accelerations in healthy fetuses from 26 to 40 weeks' gestation dem onstrated that the currently used definition of an ac celeration as ;;,15 beats/min for ;;,15 seconds is appli cable only after 30 weeks' gestational age. Decreasing the minimum amplitude to define an acceleration to 10 beats/min will have two effects on noQstress testing: (1) The length of time necessary to have a reactive nonstress test at 26 to 28 weeks will follow the same pattern as that at 30 to 40 weeks with 15 beats/min as a definition and, (2) there will be an improvement in recognition of FHR accel~rations at a basal heart rate of> 128 beats/min after 30 weeks' gestation. The data also demonstrated a maturational process in the pattern of FHR and FHR accelerations that oc curred between 28 and 30 weeks and was characterized by: (I) a decrease in basal FHR of 5 beats/min, (2) an increase in the amplitude of FHR accelerations of
4 beats/min, (3) an increase in long-term FHR vari ability, and (4) the appearance of a significant negative correlation between the mean amplitude of FHR ac celerations and the mean basal FHR. It was interesting that no further significant changes occurred in these measurements between 30 weeks and term. Current literature, between 1974 and 1984, evalu ating the predictive value for fetal outcome with ~he nonstress test has been reviewed by Devoe et a!. 1 They reported an average sensitivity of 48%, specificity of 93%, positive predictive value of 38%, and negative predictive value of 95%. A major problem in the meth odology of the studies reviewed was the criteria used for test interpretation. There were 21 different defi nitions for a "reactive" nonstress test in the 44 studies. In addition, the gestational age of fetuses undergoing testing was frequently omitted! Brown and Patrick 10 increased the length of FHR recording up to 120 minutes using the rationale that two episodes of gross fetal body movements would nor mally have been predicted during that time. 11 The in crease in observation time did not alter the negative predictive value of a reactive test defined as five or more
Patterns of FHA accelerations 747
Volume 157 Number 3
FHR accelerations of .,15 beats/min for .,15 seconds during any 20-minute interval. The present data from healthy fetuses indicated that this is an acceptable strat egy for nonstress testing of fetuses from 30 to 40 weeks' gestation, because at 60 minutes only 2. 7% of fetuses had fewer than two accelerations of., 15 beats/min for ""'15 seconds. In contrast, 65.0% of the fetuses at 26 to 28 weeks in this study had fewer than two FHR accel erations of .,15 beats/min for .,15 seconds at 60 min utes (Fig. 2, A). However, with a definition of .,10 beats/min for., 15 seconds for an acceleration in fetuses of 26 to 28 weeks' gestation, only 4.0% had fewer than two FHR accelerations at 60 minutes (Fig. 2, A). Fig. 1 indicated that in fetuses of 26 to 28 weeks' gestation 35.0% of FHR accelerations were ., 10 beats/min for .,15 seconds, which was similar to 31.7% of FHR accelerations that were ""'15 beats/min for., 15 seconds at 30 to 40 weeks' gestation. Therefore rec ognition of FHR accelerations before 30 weeks is only feasible when a minimum amplitude of., 10 beats/min for ., 15 seconds is used to define an acceleration. Dawes et al., using an improved method of fitting a baseline to normal human FHR traces," demonstrated a large rise in the numbers of accelerations after 28 weeks. 12 They also reported a significant negative cor relation between FHR variation, as expressed as the interquartile range of pulse intervals, and basal heart rate. 12 However, because of the larger scatter in the results the authors suggested that there was no material advantage to be gained by expressing the variability in FHR in terms of the basal FHR. A previous study of healthy fetuses near term showed that the mean hourly FHR was increased by 18 beats/min over 24-hour periods during times of gross fetal body movements. 13 Further, the mean daily am plitude of FHR was inversely related to the mean daily FHR. 14 Our data confirmed these findings. We also demonstrated that this same relationship was present from 30 weeks to term. This might indicate that indi vidual fetuses after 30 weeks' gestation have a set "gain" for FHR accelerations that is related to the basal heart rate. Dalton et al. 15 demonstrated in fetal lambs that si multaneous parasympathetic and 13-sympathetic block ade reduced heart rate to a greater extent in the youn ger fetuses. They suggested that there might be a rest ing cardioacceleratory drive that diminishes with age in fetal lambs. Whether or not this was related to a reduction in catecholamine sensitivity with increasing 13-sympathetic dominance during gestation remains speculative. However, this hypothesis could explain the decrease in basal FHR and the increase in long-term FHR variability reported between 26 to 28 and 30 to 32 weeks in this study. It remains to be determined if this reflected a functional maturation of the fetal au-
Y = -0.31X r = -0.56 p < 0.001
40
+ 60.77
30 20 10 0 110
120
130
140
150
160
Mean hourly basal fetal heart rote (bpm)
Fig. 5. Regression line (--) of the mean hourly amplitude of FHR accelerations was plotted in relationship to the mean hourly basal heart rate in all fetuses from 30 to 40 weeks with the 95% confidence interval ( - - - - ). A definition of ;;;. 15 beats/min for;;;. 15 seconds will detect more than 97.5% of all FHR accelerations only if the basal FHR is ,;;; 128 beats/min.
tonomic nervous system, particularly the predomi nance of the 13-sympathetic system. It is important that previous reports•· 12 • 16 and the current data suggest that the changes in basal FHR, amplitude of FHR accelerations, and long-term FHR variability occur during a critical and narrow period between 28 and 30 weeks' gestational age. Further more, no additional significant changes were measured between 30 and 40 weeks' gestational age. This sug gests that the autonomic control of FHR in healthy human fetuses might be fully mature by 30 weeks' ges tation. Data are not yet available in fetuses of less than 26 weeks' gestation. Using computerized analysis of FHR we have dem onstrated in a group of highly selected healthy fetuses that currently used criteria for nonstress testing are significantly affected by the basal FHR and gestational age. We have used instrumentation that improved heart rate recording but is not available yet for many persons. Twenty-three patients were included in the 26- to 28 week group, which represented a relatively small sam ple size, but the data suggested that using an amplitude of 10 beats/min to define an acceleration could be a strategy to increase the chance of detecting FHR ac celerations before 30 weeks' gestational age. However, this hypothesis should be clinically tested before a new definition for an acceleration is advocated. During the last 10 years there has been a dramatic decrease in neonatal deaths and morbidity in premature newborns of 26 to 28 weeks' gestational age, without any major improvements in use of FHR evaluation of fetal health at this age. Future studies evaluating antepartum FHR testing should also consider the value of adding other FHR information such as baseline heart rate, heart rate variability, and gestational age in order to increase ac curacy to predict fetal outcome with nonstress testing.
Gagnon et al.
We thank Professor G. S. Dawes and Mr. Isaac Ra poport for their help with this project and Mrs. H. Cheung and Mrs. T. Clarke for their kind technical assistance. REFERENCES 1. Devoe LD, Castillo RA, Sherline DM. The nonstress test as a diagnostic test: a critical reappraisal. AM J 0BSTET GYNECOL 1985;152:1047. 2. Thacker SB, Berkelman RL. Assessing the diagnostic ac curacy and efficacy of selected antepartum fetal surveil lance techniques. Obstet Gynecol Surv 1986;41: 121. 3. Dawes GS, Visser GHA, Goodman JDS, Redman CWG. Numerical analysis of the human fetal heart rate: the qual ity of ultrasound records. AM J 0BSTET GYNECOL 1981;141:43. 4. Dawes GS, Redman CWG, Smith JH. Improvements in the registration and analysis of fetal heart rate records at the bedside. Br J Obstet Gynaecoll985;92:317. 5. Gagnon R, Hunse C, Carmichael L, Fellows F, Patrick J. External vibroacoustic stimulation near term: fetal heart rate and heart rate variability responses. AM J OBSTET GYNECOL 1987;156:323-7. 6. Dalton KJ, Dawes GS, Patrick]. Diurnal, respiratory and other rhythms of fetal heart rate in lambs. AM J 0BSTET GYNECOL 1977;127:414. 7. Dawes GS, Houghton CRS, Redman CWG. Baseline in
September 1987 Am J Obstet Gynecol
8. 9. 10. 11. 12. 13.
14. 15. 16.
human fetal heart rate records. Br J Obstet Gynaecol 1982;89:270. Bailey BJR. Tables of the Bonferroni t statistic. J Am Statistical Assoc 1977;72:469. Kleinbaum DG, Kupper LL. Applied regression analysis and other multivariable methods. Duxbury, North Sci tuate: Duxbury Press, 1978:188. Brown R, Patrick J. The nonstress test: how long is enough. AMJ 0BSTET GYNECOL 1981;141:646. Campbell K, MacNeill I, Patrick]. Time series analysis of gross fetal body movements during the last 10 weeks of pregnancy. Ultrason Imaging 1981;3:330. Dawes GS, Houghton CRS, Redman CWG, Visser GHA. Pattern of the normal human fetal heart rate. Br J Obstet Gynaecol 1982;89:276. Patrick], Campbell K, Carmichael L, Probert C. Influence of maternal heart rate and gross fetal body movements on the daily pattern of fetal heart rate near term. AM J 0BSTET GYNECOL 1982;144:533. Patrick], Carmichael L, Chess L, Staples C. Accelerations of the human fetal heart rate at 38 to 40 weeks' gestational age. AM j 0BSTET GYNECOL 1984; 148:35. Dalton KH, Dawes GS, Patrick]. The autonomic nervous system and fetal heart rate variability. AM J 0BSTET Gv NECOL 1983;146:456. Wheeler T, Cooke E, Murrils A. Computer analysis of fetal heart rate variation during normal pregnancy. Br J Obstet Gynaecol 1979;86:186.
Riboflavin concentration in maternal and cord blood in human pregnancy Nancy Wolfert Kirshenbaum, M.D., joseph Dancis, M.D., Mortimer Levitz, Ph.D., Jean Lehanka, B.S., and Bruce K. Young, M.D. New York, New York Riboflavin concentration was measured in sera of a control population and in a series of paired maternal and cord sera. The assay technique was carefully validated and appears to be specific and reproducible. The mean riboflavin concentration in 12 apparently healthy adults was 116 ± 46 nmoi/L (SO). In 20 uneventful pregnancies the cord serum concentration was generally higher than the maternal concentration (158 ± 47 nmoi/L versus 113 ± 35 nmol/~; p = 0.001 ). The cord-to-maternal ratio in paired sera averaged 1.45 ± 0.44. There was no detectable difference in binding of riboflavin to cord and maternal serum proteins as measured by equilibrium dialysis (59.0% ± 17% versus 60.8% ± 16%). Comparison of protein binding by paired cord and mat_ernal sera yielded a ratio of 0.99 ± 0.13. The transplacental gradient of riboflavin concentration is unrelated to protein binding and is consistent with active transport by the placenta, as previously demonstrated in vitro. (AM J Oesrer GvNecoL 1987;157:748-52.)
Key words: Riboflavin concentration, maternal, cord, protein binding
From the Departments of Obstetrics and Gynecology and Pediatrics, New York University School of Medicine. Presented at the Seventh Annual Meeting of The Society ofPerinatal Obstetricians, Lake Buena Vista, Florida, February 5-7, 1987. Reprint requests:]oseph Dancis, M.D., New York University Medical Center, 550 First Ave., New York, NY 10016.
748
Riboflavin is an essential nutrient that must be deliv ered to the fetus across the placenta. The human pla centa, in vitro, actively transfers riboflavin from the maternal to the fetal circulation, establishing a gradient toward the fetus. 1' 2 The situation in vivo is less certain.
Key words: Nonstress test; fetal heart rate accelerations during third trimester; effect of gestational age on nonstress test Antepartum fetal heart rate (FHR) testing, including nonstress and contraction stress testing, has been se riously criticized 1• 2 because of a low positive predictive value, absence of strict and uniform standards for in terpretation, and the potential effect of gestational age on FHR reactivity. In addition, until recently accurate recording aQd interpretation of human FHR with Doppler ultrasound have presented m
Patients and methods Patients. informed consent was obtained from 89 healthy women with pregnancies between 26 and 40 weeks' gestational age. Eighty-three women and fetuses
From the Departments of Obstetrics and Gynecology, Physiology, and Epidemiology, University of Western Ontario, St. joseph's Hospital Research Institute. Supported by the Physicians' Services Incorporated Foundation and the Canadian Medical Research Council. Presented at the Seventh Annual Meeting of The Society of Perinatal Obstetricians, Lake Buena Vista, Florida, February 5-7, 1987. *Supported by a Fellowship from the Variety Club of Ontario. **Investigator in the Canadian Medical Research Council Group in Reproductive Biology. Reprint requests: Robert Gagnon, M.D., Department of Obstetrics and Gynecology, St. Joseph's Hospital, 268 Grosvenor St., London, Ontario N6A 4V2, Canada.
remained in good health, and six were excluded from the study. Five women were delivered of their infants before 37 weeks: Two had abruptio placentae, two had premature labor, and one was delivered of a stillborn infant at 35 weeks (4 weeks after study): All compli cations were of unknown cause. Another patient was lost to follow-up. The remaining 83 women had normal deliveries at term that confirmed gestational age and the good health of fetuses studied. The average ges tational age at delivery was 39.3 ± 0.1 weeks (SEM). The average weight of fetuses was 3350 ± 42 gm. The average 5-minute Apgar score was 9.3 ± 0.1 (mode 9; range 7 to 10). No Apgar score was less than 7 at 5 minutes. Umbilical arterial cord blood samples were obtained at birth in 58 of the 83 fetuses. The mean umbilical arterial Po 2 was 17.8 ± 1.0 mm Hg, Pco 2 was 50.0 ± 1.3 mm Hg, and pH was 7.27 ± 0.01. Ten (13.7%) women were delivered of their infants by ce sarean section: Two developed acute fetal distress in labor (5-minute Apgar scores of 8 and 10, respectively), one had cephalopelvic disproportion, one had a trans verse lie at the onset of labor, and six had repeat ce sarean sections. Eighty-one fetuses were in vertex pre sentation and two were breech at the time of delivery. Experimental design. The women received a stan dardized 800 kcal breakfast or lunch at 0800 or 1200 hours. All studies began 1 hour later and were con ducted in a quiet room with patients either sitting up right in bed or resting in a lateral recumbent position. Women were asked not to smoke for at least 4 hours before studies. Smoking or food intake by patients or observers was not permitted during the study. Women were studied continuously for 1 hour. Nine women participated in more than one study, and these were
743
744
CJ)
c:
Gagnon et al.
September 1987 Am J Obstet Gynecol
100
~
e CL>
Q;
36-40 wks • 33-35 wks • 30-32 wks e
80
(.) (.)
0
a:::
60
0
40
26-28 wks
•
I
.......
]
0 CL> 0>
c
20
0
CL>
(.)
' CL> Cl.
0 5
15
25
35
45
55
65
Amplitude of FHR accelerations (bpm) Fig. I. A cumulative histogram of 2598 FHR accelerations of ;a. 15 seconds' duration was plotted in successive increments of 5 beats/min to define the distribution of acceleration am plitudes in fetuses from 26 to 40 weeks' gestational age.
conducted in different gestational age groups. Twenty three studies were done at 26 to 28 weeks, 24 at 30 to 32 weeks, 26 at 33 to 35 weeks, and 28 at 36 to 40 weeks, for a total observation time of 101 hours. Fetal heart rate measurements. Fetal heart rate was recorded with a Hewlett-Packard 8040A (Hewlett Packard GMBH, Boeblinger, Federal Republic of Ger many) external monitor by use of a compact transducer and was analyzed on-line by a Sage II microprocessor (Sage Computer Technology, Reno, Nevada) as pre viously described.' The FHR intervals were examined by means of an artifact detection algorithm," and valid pulse intervals were averaged over 3.75 seconds. Dur ing on-line analysis a baseline was fitted to the heart rate record/ and accelerations of ;;;.J 0 beats/min for ;;;.}5 seconds above the baseline were recognized au tomatically. For each 10-minute interval the basal FHR (beats per minute) , the mean minute range (milliseconds),' and the number of FHR accelerations were calculated. The mean minute range value was calculated as the differ ence between the longest and the shortest RR intervals eyery minute and averaged for the length of observa tion. The mean minute range is a computer-derived index of long-term FHR variability.' Fetal breathing movements and gross fetal body movements were measured with a real-time ADR ul trasound scanner (Model 2130; Advanced Diagnostic Research Corp., Tempe, Arizona) and will be reported elsewhere. Data analysis. The basal FHR, the number of FHR accelerations of ;;;.JO beats/min for ;;;.15 seconds above the baseline, and amplitude of individual FHR ac celerations were determined on line in successive 10-minute intervals. The number of FHR acceler ations of ;;;.JS seconds' duration were also auto matically determined off-line for acceleration ampli-
tudes of 5 beats/min and in successive increments of 5 beats/min to a maximum amplitude of 60 beats/min. Portions of records that could not be analyzed owing to FHR signal loss of more than 30% or patient inter ruption were not included and made up 5.2% of total study time. The average signal loss of the remaining 96 hours included in this analysis was 6.9% ± 0.9%, 5.3% ± 0.6%, 5.7% ± 0.8%, and 4 .5% ± 0.5% at 26 to 28, 30 to 32, 33 to 35, and 36 to 40 weeks', respec tively. Results are presented as grouped means and stan dard errors of the mean. The significance of differ eqces in the means was examined by Bonferroni t test statistic tables to allow for multiple comparisons. 8 Stan dard regression analyses with the Pearson product moment correlations (r) were determined. Regression lines were compared for co-incidence by means of regression procedures described by Kleinbaum and Kupper! This analysis was performed with the package MINITAB (MINITAB Data Analysis Software, State College, Pennsylvania). Results Distribution of FHR accelerations. Fig. 1 is a cu mulative histogram plot of 2598 FHR accelerations of ;;;.JS seconds above the baseline in 83 fetuses between 26 and 40 weeks' gestational age. The percentage of total FHR accelerations was calculated in successive in crements of 5 beats/min to define the distribution of acceleration amplitudes. Of FHR accelerations between 30 and 40 weeks' gestation, 31.7% were ;;;.J5 beats/min for ;;;.}5 seconds. However, only 6.7% of FHR accel erations were ;;;.JS beats/min for ;;;.JS seconds at 26 to 28 weeks. The percentage of time intervals during which fewer than two FHR accelerations of ;;;.J5 beats/min for ;;;.15 seconds occurred is plotted in 10-minute increments (Fig. 2, A). From 30 to 40 weeks, 57.5 % of tracings had fewer than two FHR accelerations at 10 minutes and 2.7% at 60 minutes. However, at 26 to 28 weeks, 95.3% of recordings had fewer than two FHR accelerations at 10 minutes, 67.0% at 40 minutes, and 65.0% at 60 minutes, suggesting that most of the FHR accelerations in this group were <15 beats/min and cannot be de tected with an increase in the observation time. How ever, using a minimum amplitude of 10 beats/min for ;;;.}5 seconds to define an acceleration at 26 to 28 weeks, the length of time from the beginning of recordings until two FHR accelerations occurred was similar to that at 30 to 40 weeks (Fig. 2, A). In Fig. 2, B , it can be seen that with a decrease in the minimum amplitude to define an acceleration, there was a concomitant decrease in the percentage of 10-minute intervals during which fewer than two FHR accelerations occurred. This was observed at all ges tations from 26 to 40 weeks.
745
Patterns of FHR accelerations
Volume 157 Number 3
100
36-..0 wks
•
lQ-32 wks
• •
26-28 wks
80
150
e
ea::
33-35 .....
a:
i
140
.8
60
~
135
~
40
145
A
c:
j
A
20
130 20
0
10
0 -+-' 0
20
30
40
60
50
Time (minutes)
-+-'
Q)
0
c Q) u ..._
Q)
Q_
15
"'"
B
C:a:
2a:
100
:f .
O"l -+-'
.,, ,.,i
" u g .l!
'+
0
~~
~~ } .~
10
80
60
60
a:
55
.d E E
50
~~
45
J~
40
I~
c:c
2
B
20
::1
c
>
40 35
0
5
15
25
35
45
55
Amplitude of FHR accelerations (bpm)
Fig. 2. A, The percentage of time intervals during which fewer than two FH R accelerations of ;;.]5 beats/ min for ;;.]5 seconds occurred was plotted for all fetuses from 26 weeks to term. The percentage of time intervals during which fewer than two FHR accelerations of;;;.} 0 beats/min for ;;.]5 seconds was also plotted for fetuses from 26 to 28 weeks on the same time scale ( - - - - ). B, Cumulative percentage of I0-minute intervals dur ing which fewer than two FHR accelerations occurred was plotted in successive increments of 5 beats/min to define the minimum amplitude for an acceleration in all fetuses from 26 to 40 weeks.
Relationship between gestational age, basal FHR, and amplitude of FHR accelerations. Mean hourly ba sal FHR was 144.4 ± 1.3 beats/min at 26 to 28 weeks, which was significantly higher than at 30 to 32, 33 to 35, and 36 to 40 weeks (p < 0.05, p < 0.01, p < 0.01, respectively) (Fig. 3, A). Figure 3, B, demonstrated that there was a significant increase in the mean amplitude of FliR accelerations from 14.1 ± 0.5 beats/min at 26 to 28 weeks to 17.7 ± 0.8 beats/min at 30 to 32 weeks (p < 0.01), and this remained higher at 33 to 35 and 36 to 40 weeks (both p < 0.01) (Fig. 3, B). Finally, the 10-minute mean minute range was lower at 26 to 28 weeks than at 30 to 32, 33 to 3fi, and 36 to 40 weeks (all p < 0.01) (Fig. 3, C). It was important that there was no difference in the mean FHR, acceleration arne plitude, and mean minute range between 30 and 40 weeks' gestational age. There was a significant negative correlation between
26-28
3Q-32
33-35
36-40
Gestational age (weeks)
Fig. 3. A, The mean hourly basal FHR ( ± SEM) was plotted for fetuses at 26 to 28, 30 to 32, 33 to 35, and 36 to 40 weeks' gestation. B, The mean hourly amplitude of FHR accelera tions and, C, the mean 10-minute mean minute range (MMR) were also plotted on the same scale.
the mean hourly amplitude of FHR accelerations of ;;;.IO beats/min for ;:;;.15 seconds and the mean hourly basal FHR at 30 to 32, 33 to 35, and 36 to 40 weeks' gestation (Fig. 4, B to D), which demonstrated that within individual fetuses at 30 to 40 weeks the higher the mean hourly heart rate, the lower the height of individual FHR acceleration. No such correlation was found at 26 to 28 weeks (Fig. 4, A) . A comparison of the regression lines of the relation ship of FHR acceleration amplitudes and basal FHR at 30 to 40 weeks, by means of regression procedures described by Kleinbaum and Kupper, 9 confirmed that the regressions of Fig. 4, B to D, were coincident (slopes and intercepts were equal). Therefore these three sets of data were pooled, and Fig. 5 demonstrated the regression of FliR acceler
746 Gagnon et at.
September 1987 Am J Obstet Gynecol
35
2&-28 wks
A
30
30
25
25
20
20
... .., ..
15
8
35
30-32 wks r = -0.55 p < 0.01
.. .. . . . .· ..
15 ~
10
120
110
140
130
c
35
150
33-35 wks = -0.56 < 0.01
35
r p
30
110
160
25
25
20
20
.,.·.
15
. ....
120
130
150
160
36-40 wks = -0.66
D
r p
~
15
10
140
<
o.ooi
... . . ..·: ..
10 110
120
130
140
150
160
110
120
130
HO
150
160
Mean hourly basal fetal heart rate (bpm) Fig. 4. There was a significant negative correlation between the mean hourly amplitude of FHR accelerations and the mean hourly basal heart rate at all gestations from 30 to 40 weeks (B, C, D). No correlation was found at 26 to 28 weeks' gestation (A).
97.5 % of FHR accelerations with a basal FHR of ~ 143 beats/min in fetuses of 30 to 40 weeks' gestational age. Comment
Analysis of the distributior) of FHR accelerations in healthy fetuses from 26 to 40 weeks' gestation dem onstrated that the currently used definition of an ac celeration as ;;,15 beats/min for ;;,15 seconds is appli cable only after 30 weeks' gestational age. Decreasing the minimum amplitude to define an acceleration to 10 beats/min will have two effects on noQstress testing: (1) The length of time necessary to have a reactive nonstress test at 26 to 28 weeks will follow the same pattern as that at 30 to 40 weeks with 15 beats/min as a definition and, (2) there will be an improvement in recognition of FHR accel~rations at a basal heart rate of> 128 beats/min after 30 weeks' gestation. The data also demonstrated a maturational process in the pattern of FHR and FHR accelerations that oc curred between 28 and 30 weeks and was characterized by: (I) a decrease in basal FHR of 5 beats/min, (2) an increase in the amplitude of FHR accelerations of
4 beats/min, (3) an increase in long-term FHR vari ability, and (4) the appearance of a significant negative correlation between the mean amplitude of FHR ac celerations and the mean basal FHR. It was interesting that no further significant changes occurred in these measurements between 30 weeks and term. Current literature, between 1974 and 1984, evalu ating the predictive value for fetal outcome with ~he nonstress test has been reviewed by Devoe et a!. 1 They reported an average sensitivity of 48%, specificity of 93%, positive predictive value of 38%, and negative predictive value of 95%. A major problem in the meth odology of the studies reviewed was the criteria used for test interpretation. There were 21 different defi nitions for a "reactive" nonstress test in the 44 studies. In addition, the gestational age of fetuses undergoing testing was frequently omitted! Brown and Patrick 10 increased the length of FHR recording up to 120 minutes using the rationale that two episodes of gross fetal body movements would nor mally have been predicted during that time. 11 The in crease in observation time did not alter the negative predictive value of a reactive test defined as five or more
Patterns of FHA accelerations 747
Volume 157 Number 3
FHR accelerations of .,15 beats/min for .,15 seconds during any 20-minute interval. The present data from healthy fetuses indicated that this is an acceptable strat egy for nonstress testing of fetuses from 30 to 40 weeks' gestation, because at 60 minutes only 2. 7% of fetuses had fewer than two accelerations of., 15 beats/min for ""'15 seconds. In contrast, 65.0% of the fetuses at 26 to 28 weeks in this study had fewer than two FHR accel erations of .,15 beats/min for .,15 seconds at 60 min utes (Fig. 2, A). However, with a definition of .,10 beats/min for., 15 seconds for an acceleration in fetuses of 26 to 28 weeks' gestation, only 4.0% had fewer than two FHR accelerations at 60 minutes (Fig. 2, A). Fig. 1 indicated that in fetuses of 26 to 28 weeks' gestation 35.0% of FHR accelerations were ., 10 beats/min for .,15 seconds, which was similar to 31.7% of FHR accelerations that were ""'15 beats/min for., 15 seconds at 30 to 40 weeks' gestation. Therefore rec ognition of FHR accelerations before 30 weeks is only feasible when a minimum amplitude of., 10 beats/min for ., 15 seconds is used to define an acceleration. Dawes et al., using an improved method of fitting a baseline to normal human FHR traces," demonstrated a large rise in the numbers of accelerations after 28 weeks. 12 They also reported a significant negative cor relation between FHR variation, as expressed as the interquartile range of pulse intervals, and basal heart rate. 12 However, because of the larger scatter in the results the authors suggested that there was no material advantage to be gained by expressing the variability in FHR in terms of the basal FHR. A previous study of healthy fetuses near term showed that the mean hourly FHR was increased by 18 beats/min over 24-hour periods during times of gross fetal body movements. 13 Further, the mean daily am plitude of FHR was inversely related to the mean daily FHR. 14 Our data confirmed these findings. We also demonstrated that this same relationship was present from 30 weeks to term. This might indicate that indi vidual fetuses after 30 weeks' gestation have a set "gain" for FHR accelerations that is related to the basal heart rate. Dalton et al. 15 demonstrated in fetal lambs that si multaneous parasympathetic and 13-sympathetic block ade reduced heart rate to a greater extent in the youn ger fetuses. They suggested that there might be a rest ing cardioacceleratory drive that diminishes with age in fetal lambs. Whether or not this was related to a reduction in catecholamine sensitivity with increasing 13-sympathetic dominance during gestation remains speculative. However, this hypothesis could explain the decrease in basal FHR and the increase in long-term FHR variability reported between 26 to 28 and 30 to 32 weeks in this study. It remains to be determined if this reflected a functional maturation of the fetal au-
Y = -0.31X r = -0.56 p < 0.001
40
+ 60.77
30 20 10 0 110
120
130
140
150
160
Mean hourly basal fetal heart rote (bpm)
Fig. 5. Regression line (--) of the mean hourly amplitude of FHR accelerations was plotted in relationship to the mean hourly basal heart rate in all fetuses from 30 to 40 weeks with the 95% confidence interval ( - - - - ). A definition of ;;;. 15 beats/min for;;;. 15 seconds will detect more than 97.5% of all FHR accelerations only if the basal FHR is ,;;; 128 beats/min.
tonomic nervous system, particularly the predomi nance of the 13-sympathetic system. It is important that previous reports•· 12 • 16 and the current data suggest that the changes in basal FHR, amplitude of FHR accelerations, and long-term FHR variability occur during a critical and narrow period between 28 and 30 weeks' gestational age. Further more, no additional significant changes were measured between 30 and 40 weeks' gestational age. This sug gests that the autonomic control of FHR in healthy human fetuses might be fully mature by 30 weeks' ges tation. Data are not yet available in fetuses of less than 26 weeks' gestation. Using computerized analysis of FHR we have dem onstrated in a group of highly selected healthy fetuses that currently used criteria for nonstress testing are significantly affected by the basal FHR and gestational age. We have used instrumentation that improved heart rate recording but is not available yet for many persons. Twenty-three patients were included in the 26- to 28 week group, which represented a relatively small sam ple size, but the data suggested that using an amplitude of 10 beats/min to define an acceleration could be a strategy to increase the chance of detecting FHR ac celerations before 30 weeks' gestational age. However, this hypothesis should be clinically tested before a new definition for an acceleration is advocated. During the last 10 years there has been a dramatic decrease in neonatal deaths and morbidity in premature newborns of 26 to 28 weeks' gestational age, without any major improvements in use of FHR evaluation of fetal health at this age. Future studies evaluating antepartum FHR testing should also consider the value of adding other FHR information such as baseline heart rate, heart rate variability, and gestational age in order to increase ac curacy to predict fetal outcome with nonstress testing.
Gagnon et al.
We thank Professor G. S. Dawes and Mr. Isaac Ra poport for their help with this project and Mrs. H. Cheung and Mrs. T. Clarke for their kind technical assistance. REFERENCES 1. Devoe LD, Castillo RA, Sherline DM. The nonstress test as a diagnostic test: a critical reappraisal. AM J 0BSTET GYNECOL 1985;152:1047. 2. Thacker SB, Berkelman RL. Assessing the diagnostic ac curacy and efficacy of selected antepartum fetal surveil lance techniques. Obstet Gynecol Surv 1986;41: 121. 3. Dawes GS, Visser GHA, Goodman JDS, Redman CWG. Numerical analysis of the human fetal heart rate: the qual ity of ultrasound records. AM J 0BSTET GYNECOL 1981;141:43. 4. Dawes GS, Redman CWG, Smith JH. Improvements in the registration and analysis of fetal heart rate records at the bedside. Br J Obstet Gynaecoll985;92:317. 5. Gagnon R, Hunse C, Carmichael L, Fellows F, Patrick J. External vibroacoustic stimulation near term: fetal heart rate and heart rate variability responses. AM J OBSTET GYNECOL 1987;156:323-7. 6. Dalton KJ, Dawes GS, Patrick]. Diurnal, respiratory and other rhythms of fetal heart rate in lambs. AM J 0BSTET GYNECOL 1977;127:414. 7. Dawes GS, Houghton CRS, Redman CWG. Baseline in
September 1987 Am J Obstet Gynecol
8. 9. 10. 11. 12. 13.
14. 15. 16.
human fetal heart rate records. Br J Obstet Gynaecol 1982;89:270. Bailey BJR. Tables of the Bonferroni t statistic. J Am Statistical Assoc 1977;72:469. Kleinbaum DG, Kupper LL. Applied regression analysis and other multivariable methods. Duxbury, North Sci tuate: Duxbury Press, 1978:188. Brown R, Patrick J. The nonstress test: how long is enough. AMJ 0BSTET GYNECOL 1981;141:646. Campbell K, MacNeill I, Patrick]. Time series analysis of gross fetal body movements during the last 10 weeks of pregnancy. Ultrason Imaging 1981;3:330. Dawes GS, Houghton CRS, Redman CWG, Visser GHA. Pattern of the normal human fetal heart rate. Br J Obstet Gynaecol 1982;89:276. Patrick], Campbell K, Carmichael L, Probert C. Influence of maternal heart rate and gross fetal body movements on the daily pattern of fetal heart rate near term. AM J 0BSTET GYNECOL 1982;144:533. Patrick], Carmichael L, Chess L, Staples C. Accelerations of the human fetal heart rate at 38 to 40 weeks' gestational age. AM j 0BSTET GYNECOL 1984; 148:35. Dalton KH, Dawes GS, Patrick]. The autonomic nervous system and fetal heart rate variability. AM J 0BSTET Gv NECOL 1983;146:456. Wheeler T, Cooke E, Murrils A. Computer analysis of fetal heart rate variation during normal pregnancy. Br J Obstet Gynaecol 1979;86:186.
Riboflavin concentration in maternal and cord blood in human pregnancy Nancy Wolfert Kirshenbaum, M.D., joseph Dancis, M.D., Mortimer Levitz, Ph.D., Jean Lehanka, B.S., and Bruce K. Young, M.D. New York, New York Riboflavin concentration was measured in sera of a control population and in a series of paired maternal and cord sera. The assay technique was carefully validated and appears to be specific and reproducible. The mean riboflavin concentration in 12 apparently healthy adults was 116 ± 46 nmoi/L (SO). In 20 uneventful pregnancies the cord serum concentration was generally higher than the maternal concentration (158 ± 47 nmoi/L versus 113 ± 35 nmol/~; p = 0.001 ). The cord-to-maternal ratio in paired sera averaged 1.45 ± 0.44. There was no detectable difference in binding of riboflavin to cord and maternal serum proteins as measured by equilibrium dialysis (59.0% ± 17% versus 60.8% ± 16%). Comparison of protein binding by paired cord and mat_ernal sera yielded a ratio of 0.99 ± 0.13. The transplacental gradient of riboflavin concentration is unrelated to protein binding and is consistent with active transport by the placenta, as previously demonstrated in vitro. (AM J Oesrer GvNecoL 1987;157:748-52.)
Key words: Riboflavin concentration, maternal, cord, protein binding
From the Departments of Obstetrics and Gynecology and Pediatrics, New York University School of Medicine. Presented at the Seventh Annual Meeting of The Society ofPerinatal Obstetricians, Lake Buena Vista, Florida, February 5-7, 1987. Reprint requests:]oseph Dancis, M.D., New York University Medical Center, 550 First Ave., New York, NY 10016.
748
Riboflavin is an essential nutrient that must be deliv ered to the fetus across the placenta. The human pla centa, in vitro, actively transfers riboflavin from the maternal to the fetal circulation, establishing a gradient toward the fetus. 1' 2 The situation in vivo is less certain.