- Email: [email protected]
The effect of different sampling intervals on the measurement of intrapartum fetal heart rate variability
The Effect of Different Sampling Intervals Measurement of Intrapartum Fetal Heart Rate Variability M. A. WKCOX, MD, W. WANG, PhD, D. S. SAHOTA, T. K. H. CHUNG, AND A. M. Z. CHANG, PhD Objective: To test the hypothesis that increasing the sampling interval affects the intrapartum fetal heart rate (FHR) variability measurement. Mefhods: Fetal electrocardiograms were obtained from women in labor. Using the peak of the fetal R wave, the R-R interval and FHR were calculated on a beat-to-beat basis. Retrospectively, the original data were repartitioned using different intervals (2-900 seconds) to generate a window of measurement (epoch). The mean value for each epoch and the last FHR in that epoch (epochal value) were compared with published animal and human data. Errors were quantified by comparing the epochal and mean values for each epoch. Fetal heart rate variability between epochs and within each epoch was compared. Results: Fetal heart rate and R-R interval were measured in 146 cases. The FHR had a normal distribution (mean 140.1 beats per minute, f standard deviation [SD] 15.6, skew -0.07), but its inverse, the R-R interval, was not normally distributed (mean 432 milliseconds, f SD 52.4, skew 1.78). Using a single value for an epoch duration of 2 seconds resulted in an error that was similar to the within-epoch variability (* SD of 2.2 beats per minute difference between mean and epochal value compared to f SD of 2 beats per minute within epoch) but which increased with epoch duration. Conclusion: An epoch duration of 2 seconds and a single sampled value within this period may be appropriate for measurement of both medium and long-term variability in any computerized intrapartum FHR interpretation system. Fetal heart rate (not R-R interval, because of its normal distribution) should be used to design such a computerized
From the Department qf Obstetrics and Gynaecology, The Chinese University ofHong Kong, Prince of Wales Hospital, Shatin, Hong Kong; and the Department qf Obstetrics and Gynaecology, The University of Nottingham, Nottingham, United Kingdom. We acknowledge the support @The British Council in the development of academic links between the Chinese University 4 Hong Kong and Nottingham University, and Prqfessor E. M. Symondsfor the use of the Nottingham Fetal Electrocardiogram Analyzer.
VOL.
89,NO.4,APRIL
1997
on the
PhD, R. R. DAWKINS,
system. (Obstet Gynecol 1997;89:577-80. 0 1997 by The American College of Obstetricians and Gynecologists.)
The use of the cardiotocograph for antepartum and intrapartum fetal surveillance is well established. However, there is substantial inter- and intra-observer variation in its visual interpretation.‘” Therefore, considerable effort has been expended to introduce computerized models that give, at least, a repeatable interpretation. Much of the work hasconcentrated on animal or antenatal data, however, and it is uncertain how these models can be applied to the human intrapartum situation. Cardiotocographic changes occur more rapidly during labor than in the relatively quiet antenatal period, and the interpretation of these changes needs to be made as they are happening, not after the event. Therefore, online intrapartum computerized interpretations would be more useful if developed specifically for this purpose, and not just as modifications of existing antenatal models4 Most computerized interpretation systems use fixedinterval fetal heart rate (FHR) sampling instead of recording each heartbeat. The advantages of using a constant sampling interval is the easewith which data can be handled by the hardware and software for analysis and statistical assessment.The disadvantage is that not all possible data are included. This work is important if computerized assessmentis to be introduced into routine practice. If FHR data are to be analyzed, individual values must first be obtained, and this in turn can be done only by sampling the FHR over time. Therefore, computerized measurements of variability can be used only after the effects of differing sampling intervals have been clarified. Using different sampling intervals, we tested the hypothesis that in-
0029-7844/97/$17.00 PI1 SOO29-7844(97)00060-4
577
creasing sampling interval affects the measurements of variability of intrapartum FHR by describing and quantifying any bias that may be introduced.
Materials and Methods Fetal heart rate data were recorded in women in labor at Prince of Wales Hospital, Hong Kong. We recruited patients for the study after the decision to attach a fetal scalp electrode was made for clinical indications. Management of the labor was not affected in any way by this study, and no women were excluded. Prior approval was obtained from the Hospital Ethics Committee, and written patient consent also was obtained. The fetal electrocardiogram was obtained from a single spiral electrode applied to the fetal scalp, electrically isolated with the Corometrics 115 or 116 cardiotocograph monitor (Corometrics, Wallingford, CT), then assessed with the Nottingham Fetal Electrocardiogram Analyzer.5 The fetal electrocardiogram signal was digitized at a sampling frequency of 500 Hz, allowing R-R peak-to-peak intervals to be determined to an accuracy of +-1 millisecond. The FHR and R-R intervals were then stored on the analyzer’s hard disk in time-stamped files. The data files were partitioned retrospectively in intervals of increasing duration, from beat to beat up to a period of 20 seconds in 2-second increments, then at larger intervals up to a maximum duration of 900 seconds. Each of the data blocks so produced was termed an epoch. The mean of all the values within each epoch and the last value of FHR within that epoch (epochal value) were selected for analysis. The goodness of fit to the normal distribution of all recorded FHR and R-R intervals was analyzed with the Kolmogorov-Smirnov test6 The effect of epoch duration on the proportion of blocks containing no data was studied. Heart rate variability was calculated with a hierarchical analysis of variance, separating total variance into that between epochs and within each epoch. The variability of R-R intervals in milliseconds as well as FHR in beats per minute were calculated. The mean of standard deviations (SDS) of R-R intervals for all epochs for each epoch duration was plotted against epoch duration, and this was compared with published animal and human data.7,s Error was quantified by comparing the mean and epochal values for each epoch. These calculations allow the comparison of the intrapartum FHR data obtained from this study with that reported previously.
Results Fetal heart rate data were obtained from 146 women in labor, all of Chinese extraction, producing 4,624,750 R-R
578
Wilcox
et al
FHR Variability and Epoch Duration
=
600
8;; ‘5; ;
500 400
s = 300 p! IA 200 100
0
8 2 zii 2 8 % 5: $ 22 8 Recorded
Figure
1. Distribution
fetarheart ranges(beats of intrapartum
rat: val;es i: f&bear per minute) fetal heart rate values recorded.
interval measurements. The mean (2 SD) age of these women was 29.7 ? 5 years. One hundred one women (69%) were nulliparous, and nine (6%) had three or more live births. The infants when born had a mean (2 SD) birth weight of 3278 -+ 575 g, gestational age at delivery of 279 i 12.4 days, and umbilical cord arterial blood pH of 7.26 ? 0.07. Thirty-seven (25.3%) fetuses were delivered by cesarean, and 34 (23.3%) were delivered vaginally. The cord arterial pH was less than 7.10 in three cases, and the base deficit exceeded 10 mmol/L in six cases. The distribution of recorded FHRs is shown in Figure 1. The mean + SD was 140.1 + 15.6 beats per minute (skew -0.07). The analysis of R-R intervals revealed a distribution with a mean of 432 ? 52.4 milliseconds (skew 1.78). The number of epochs containing no FHR values decreased as the duration of each epoch increased. When a 2-second duration was used, 2.54% of epochs contained no data. This decreased to 0.06% when a 20-second epoch was used. Figure 2 illustrates the relationship between epoch duration and measurement of variability (SD of FHRs), indicating that all measurements of variability increase with epoch duration. Figure 3 indicates the cumulative distribution of beat-to-beat differences in R-R intervals obtained in this study and that of two previous studies,7,B showing that beat-to-beat differences obtained during labor in this study were much greater than those obtained from the previous antepartum or animal models. Figure 4 illustrates the relationship between epoch duration and the mean of all epoch R-R intervals with that duration, indicating the linear relationship described by Dalton et al7 and the sigmoidal relationship found in this study.
Obstetrics 6 Gynecology
-SD
within
-l\lean
epochs
of absolute
@?5D
of absolute
-SD
of difference
difference difference
between between
between
epochal epochal
mean and epochal
values values value
Observation
epoch
duration(seconds)
Figure 4. Comparison of the mean of the standard deviation of R-R intervals found in this study with another study. Our own data support a sigmoidal relationship, unlike Dalton et aL7 who reported a linear relationship for epoch durations longer than 15 seconds.
5 10 20 15 Obervation epoch duration(seconds)
0
Figure 2. Graphic display of standard deviation (SD) obtained as epoch duration increases. Analysis of variance (SD*) was used to compare differences within and between different-sized epochs and their mean and epochal values.
Discussion For various reasons, previous computerized interpretations of the cardiotocograph have used different epoch durations, including 3.75 seconds,” 2 seconds,” and 1 second.” Although the effects of increasing epoch du-
2 zi g
80
5 e g
60
.-B f;; 3
40
:
E
70 50
30 20
0
2
4 Difference
6
6
Figure 3. Comparison of cumulative R-R interval difference this study and that from sheep and human subjects.
VOL.
89,
NO.
4, APRIL
1997
10
(milliseconds)
found
in
rations on measurement of variability greater than 15 seconds have been evaluated, there are few data when shorter periods are used. The Nottingham Fetal Electrocardiogram Analyzer5 allows the relationship between sampling interval and variability to be studied with epoch durations less than 15 seconds, directly from the R-R intervals of the fetal electrocardiogram, free from the electronic interference of maternal movement and from averaging or autocorrelation techniques. Differences in beat-to-beat intervals were greater in our study than in previous studies (Figure 3).7,s However, there are marked differences in the methodologies used, making comparisons difficult. Dalton et al7 used invasive techniques to study anesthetized fetal lambs under a variety of experimental conditions. Wheeler et al* used the transabdominal fetal electrocardiogram in normal human antenatal subjects. In our study, the differences observed in successive R-R intervals more closely resemble the human data presented by Wheeler et al,s but the differences may reflect physiologic differences between antenatal and intrapartum conditions. We found that the positive relationship between epoch duration and mean SD of all R-R intervals for a given epoch duration was most marked when shorter intervals were used. Past 120 seconds, there was little change in the mean SD (Figure 4). In previous studies’s8 the logarithm of epoch duration was related linearly, and this was used to justify an extrapolation’ toward a sampling interval of 3.75 seconds. However, our findings suggest that the relationship is sigmoidal when a wider range of epoch durations are used, and the extrapolation reported may not be appropriate. Although there was some change in variability for epoch
Wilcox
et al
FHR
Variability
and Epoch
Duration
579
durations greater than 120 seconds, these changes were much less than those found for shorter durations. The greater variability found in longer epochs may be due to the inclusion of periodic changes, such as accelerations and decelerations. Therefore, systems developed using data from animals or antepartum women must be modified before their application to intrapartum analysis. We observed that the longer the epoch duration, the more likely the epoch is to contain some meaningful data. However, as epoch duration increases, the chance of including an acceleration or deceleration also increases, in turn increasing the variability within the epoch (Figure 2). For this reason, the absolute difference in heart rate between successive epochs appears to increase with increasing epoch duration. Therefore, the selection of a suitable epoch duration is a compromise between the probability of getting an epoch with valid data and avoiding a misleading increased measurement of variability due to presence of accelerations or decelerations. The representative FHR for a given epoch (epochal value) may be either its mean value or a single sampled value from within that epoch, the appropriateness of either being dependent on the epoch duration. As epoch duration increases, so does the SD of FHR within that epoch (Figure 2), and a single-sampled value will be less representative of that epoch. Therefore, within the confines of any system, the shortest possible epoch duration should be chosen. We suggest a 2-second duration be adopted because anything shorter will result in an excessive number of epochs containing no valid data. At 2 seconds, the error in using a singlesampled value is similar to the within-epoch variability (-+ SD of 2.2 beats per minute between epoch compared to -C SD of 2 beats per minute within epoch), whereas for durations of 4 seconds, the error increases (i. SD 3.5 beats per minute compared to + SD 2.8 beats per minute). A given change in heart rate will have a varying affect on the R-R interval, depending on the initial rate. An increase of 10 beats per minute at 120 beats per minute involves an R-R interval change of 38.5 milliseconds, whereas at 160 beats per minute, the change is only 22.5 milliseconds. Our study on distribution suggests that because FHR is almost normally distributed, it is an easier variable to handle statistically, and we recommend the use of FHR rather than R-R interval in the computerized interpretation of the cardiotocogram. In conclusion, episodic sampling of the FHR offers advantages in the intrapartum computerized interpre-
580
Wilcox
et al
FHR Variability and Epoch Duration
tation of the cardiotocograph. We suggest an epoch duration of 2 seconds and the adoptation of a single sampled value within this period for measurement of both medium and long-term variability. Because FHR is normally distributed, we suggest it be used instead of R-R intervals. Finally, intrapartum FHR differs from antenatal FHR, and we suggest that this difference be considered when developing computerized systems for interpretation of intrapartum cardiotocograph.
Rt$erences 1 Trimbos JB, Keirse MJNC. Observer variability in assessment of antepartum cardiotocograms. Br J Obstet Gynaecol 1978;85:900-6. 2 Lotgering FK, Wallenburg HCS, Schouten HJA. Interobserver and intraobserver variation in the assessment of antepartum cardiotocograms. Am J Obstet Gynecol 1982;144:701-5. 3 Donker DK, van Geijn HP, Hasman A. Interobserver variation in the assessment of FHR recordings. Eur J Obstet Gynecol Reprod Biol 1993;52:21-8. 4. Pello LC, Rosevear SK, Dawes GS, Moulden M, Redman CWG. Computerized FHR analysis in labor. Obstet Gynecol 1991;78:60210. 5. Mohajer Ml’, Sahota DS, Reed NN, Chang AMZ, Symonds EM, James DK. Cumulative changes in the fetal electrocardiogram and biochemical indices of fetal hypoxia. Eur J Obstet Gynecol Reprod Biol 1994;55:63-70. 6. Siegel S, Castellan NJ. Nonparametric statistics for the behavioural sciences. 2nd ed. New York: McGraw-Hill, 198851-5. 7. Dalton KJ, Dawes GS, Patrick JE. Diurnal, respiratory, and other rhythms of FHR in lambs. Am J Obstet Gynecol 1977;127:414-24. 8. Wheeler T, Cooke E, Mm-rills A. Computer analysis of FHR variation during normal pregnancy. Br J Obstet Gynaecol 1979;86: 186-97. 9. Dawes GS, Moulden M, Redman CWG. Criteria for the design of FHR analysis systems. Int J Biomed Comput 1990;25:287-94. 10. Maeda K. Computerized analysis of cardiotocograms and fetal movements. Baillieres Clin Obstet Gynaecol 1990;4:797-813. 11. Stigsby B, Nielsen PV, Olesen MB, Jaszczak I’, Larsen JF, Lenstrup C. Computer description of the cardiotocogram. 1. The computer program. Acta Obstet Gynecol Stand 1982;109:76-8.
Address
reprint
requests
to:
D. S. Sahota, PhD Department qf Obstetrics and Gynecology Prince qf Wales Hospital Shatin, NT Hong Kong
Receiued August 26, 1996. Received in revisedform December 6, 1996. Accepted Janua y 16, 1997. Copyright 0 1997 by The American College of Obstetricians Gynecologists. Published by Elsevier Science Inc.
and
Obstetrics b Gynecology
From the Department qf Obstetrics and Gynaecology, The Chinese University ofHong Kong, Prince of Wales Hospital, Shatin, Hong Kong; and the Department qf Obstetrics and Gynaecology, The University of Nottingham, Nottingham, United Kingdom. We acknowledge the support @The British Council in the development of academic links between the Chinese University 4 Hong Kong and Nottingham University, and Prqfessor E. M. Symondsfor the use of the Nottingham Fetal Electrocardiogram Analyzer.
VOL.
89,NO.4,APRIL
1997
on the
PhD, R. R. DAWKINS,
system. (Obstet Gynecol 1997;89:577-80. 0 1997 by The American College of Obstetricians and Gynecologists.)
The use of the cardiotocograph for antepartum and intrapartum fetal surveillance is well established. However, there is substantial inter- and intra-observer variation in its visual interpretation.‘” Therefore, considerable effort has been expended to introduce computerized models that give, at least, a repeatable interpretation. Much of the work hasconcentrated on animal or antenatal data, however, and it is uncertain how these models can be applied to the human intrapartum situation. Cardiotocographic changes occur more rapidly during labor than in the relatively quiet antenatal period, and the interpretation of these changes needs to be made as they are happening, not after the event. Therefore, online intrapartum computerized interpretations would be more useful if developed specifically for this purpose, and not just as modifications of existing antenatal models4 Most computerized interpretation systems use fixedinterval fetal heart rate (FHR) sampling instead of recording each heartbeat. The advantages of using a constant sampling interval is the easewith which data can be handled by the hardware and software for analysis and statistical assessment.The disadvantage is that not all possible data are included. This work is important if computerized assessmentis to be introduced into routine practice. If FHR data are to be analyzed, individual values must first be obtained, and this in turn can be done only by sampling the FHR over time. Therefore, computerized measurements of variability can be used only after the effects of differing sampling intervals have been clarified. Using different sampling intervals, we tested the hypothesis that in-
0029-7844/97/$17.00 PI1 SOO29-7844(97)00060-4
577
creasing sampling interval affects the measurements of variability of intrapartum FHR by describing and quantifying any bias that may be introduced.
Materials and Methods Fetal heart rate data were recorded in women in labor at Prince of Wales Hospital, Hong Kong. We recruited patients for the study after the decision to attach a fetal scalp electrode was made for clinical indications. Management of the labor was not affected in any way by this study, and no women were excluded. Prior approval was obtained from the Hospital Ethics Committee, and written patient consent also was obtained. The fetal electrocardiogram was obtained from a single spiral electrode applied to the fetal scalp, electrically isolated with the Corometrics 115 or 116 cardiotocograph monitor (Corometrics, Wallingford, CT), then assessed with the Nottingham Fetal Electrocardiogram Analyzer.5 The fetal electrocardiogram signal was digitized at a sampling frequency of 500 Hz, allowing R-R peak-to-peak intervals to be determined to an accuracy of +-1 millisecond. The FHR and R-R intervals were then stored on the analyzer’s hard disk in time-stamped files. The data files were partitioned retrospectively in intervals of increasing duration, from beat to beat up to a period of 20 seconds in 2-second increments, then at larger intervals up to a maximum duration of 900 seconds. Each of the data blocks so produced was termed an epoch. The mean of all the values within each epoch and the last value of FHR within that epoch (epochal value) were selected for analysis. The goodness of fit to the normal distribution of all recorded FHR and R-R intervals was analyzed with the Kolmogorov-Smirnov test6 The effect of epoch duration on the proportion of blocks containing no data was studied. Heart rate variability was calculated with a hierarchical analysis of variance, separating total variance into that between epochs and within each epoch. The variability of R-R intervals in milliseconds as well as FHR in beats per minute were calculated. The mean of standard deviations (SDS) of R-R intervals for all epochs for each epoch duration was plotted against epoch duration, and this was compared with published animal and human data.7,s Error was quantified by comparing the mean and epochal values for each epoch. These calculations allow the comparison of the intrapartum FHR data obtained from this study with that reported previously.
Results Fetal heart rate data were obtained from 146 women in labor, all of Chinese extraction, producing 4,624,750 R-R
578
Wilcox
et al
FHR Variability and Epoch Duration
=
600
8;; ‘5; ;
500 400
s = 300 p! IA 200 100
0
8 2 zii 2 8 % 5: $ 22 8 Recorded
Figure
1. Distribution
fetarheart ranges(beats of intrapartum
rat: val;es i: f&bear per minute) fetal heart rate values recorded.
interval measurements. The mean (2 SD) age of these women was 29.7 ? 5 years. One hundred one women (69%) were nulliparous, and nine (6%) had three or more live births. The infants when born had a mean (2 SD) birth weight of 3278 -+ 575 g, gestational age at delivery of 279 i 12.4 days, and umbilical cord arterial blood pH of 7.26 ? 0.07. Thirty-seven (25.3%) fetuses were delivered by cesarean, and 34 (23.3%) were delivered vaginally. The cord arterial pH was less than 7.10 in three cases, and the base deficit exceeded 10 mmol/L in six cases. The distribution of recorded FHRs is shown in Figure 1. The mean + SD was 140.1 + 15.6 beats per minute (skew -0.07). The analysis of R-R intervals revealed a distribution with a mean of 432 ? 52.4 milliseconds (skew 1.78). The number of epochs containing no FHR values decreased as the duration of each epoch increased. When a 2-second duration was used, 2.54% of epochs contained no data. This decreased to 0.06% when a 20-second epoch was used. Figure 2 illustrates the relationship between epoch duration and measurement of variability (SD of FHRs), indicating that all measurements of variability increase with epoch duration. Figure 3 indicates the cumulative distribution of beat-to-beat differences in R-R intervals obtained in this study and that of two previous studies,7,B showing that beat-to-beat differences obtained during labor in this study were much greater than those obtained from the previous antepartum or animal models. Figure 4 illustrates the relationship between epoch duration and the mean of all epoch R-R intervals with that duration, indicating the linear relationship described by Dalton et al7 and the sigmoidal relationship found in this study.
Obstetrics 6 Gynecology
-SD
within
-l\lean
epochs
of absolute
@?5D
of absolute
-SD
of difference
difference difference
between between
between
epochal epochal
mean and epochal
values values value
Observation
epoch
duration(seconds)
Figure 4. Comparison of the mean of the standard deviation of R-R intervals found in this study with another study. Our own data support a sigmoidal relationship, unlike Dalton et aL7 who reported a linear relationship for epoch durations longer than 15 seconds.
5 10 20 15 Obervation epoch duration(seconds)
0
Figure 2. Graphic display of standard deviation (SD) obtained as epoch duration increases. Analysis of variance (SD*) was used to compare differences within and between different-sized epochs and their mean and epochal values.
Discussion For various reasons, previous computerized interpretations of the cardiotocograph have used different epoch durations, including 3.75 seconds,” 2 seconds,” and 1 second.” Although the effects of increasing epoch du-
2 zi g
80
5 e g
60
.-B f;; 3
40
:
E
70 50
30 20
0
2
4 Difference
6
6
Figure 3. Comparison of cumulative R-R interval difference this study and that from sheep and human subjects.
VOL.
89,
NO.
4, APRIL
1997
10
(milliseconds)
found
in
rations on measurement of variability greater than 15 seconds have been evaluated, there are few data when shorter periods are used. The Nottingham Fetal Electrocardiogram Analyzer5 allows the relationship between sampling interval and variability to be studied with epoch durations less than 15 seconds, directly from the R-R intervals of the fetal electrocardiogram, free from the electronic interference of maternal movement and from averaging or autocorrelation techniques. Differences in beat-to-beat intervals were greater in our study than in previous studies (Figure 3).7,s However, there are marked differences in the methodologies used, making comparisons difficult. Dalton et al7 used invasive techniques to study anesthetized fetal lambs under a variety of experimental conditions. Wheeler et al* used the transabdominal fetal electrocardiogram in normal human antenatal subjects. In our study, the differences observed in successive R-R intervals more closely resemble the human data presented by Wheeler et al,s but the differences may reflect physiologic differences between antenatal and intrapartum conditions. We found that the positive relationship between epoch duration and mean SD of all R-R intervals for a given epoch duration was most marked when shorter intervals were used. Past 120 seconds, there was little change in the mean SD (Figure 4). In previous studies’s8 the logarithm of epoch duration was related linearly, and this was used to justify an extrapolation’ toward a sampling interval of 3.75 seconds. However, our findings suggest that the relationship is sigmoidal when a wider range of epoch durations are used, and the extrapolation reported may not be appropriate. Although there was some change in variability for epoch
Wilcox
et al
FHR
Variability
and Epoch
Duration
579
durations greater than 120 seconds, these changes were much less than those found for shorter durations. The greater variability found in longer epochs may be due to the inclusion of periodic changes, such as accelerations and decelerations. Therefore, systems developed using data from animals or antepartum women must be modified before their application to intrapartum analysis. We observed that the longer the epoch duration, the more likely the epoch is to contain some meaningful data. However, as epoch duration increases, the chance of including an acceleration or deceleration also increases, in turn increasing the variability within the epoch (Figure 2). For this reason, the absolute difference in heart rate between successive epochs appears to increase with increasing epoch duration. Therefore, the selection of a suitable epoch duration is a compromise between the probability of getting an epoch with valid data and avoiding a misleading increased measurement of variability due to presence of accelerations or decelerations. The representative FHR for a given epoch (epochal value) may be either its mean value or a single sampled value from within that epoch, the appropriateness of either being dependent on the epoch duration. As epoch duration increases, so does the SD of FHR within that epoch (Figure 2), and a single-sampled value will be less representative of that epoch. Therefore, within the confines of any system, the shortest possible epoch duration should be chosen. We suggest a 2-second duration be adopted because anything shorter will result in an excessive number of epochs containing no valid data. At 2 seconds, the error in using a singlesampled value is similar to the within-epoch variability (-+ SD of 2.2 beats per minute between epoch compared to -C SD of 2 beats per minute within epoch), whereas for durations of 4 seconds, the error increases (i. SD 3.5 beats per minute compared to + SD 2.8 beats per minute). A given change in heart rate will have a varying affect on the R-R interval, depending on the initial rate. An increase of 10 beats per minute at 120 beats per minute involves an R-R interval change of 38.5 milliseconds, whereas at 160 beats per minute, the change is only 22.5 milliseconds. Our study on distribution suggests that because FHR is almost normally distributed, it is an easier variable to handle statistically, and we recommend the use of FHR rather than R-R interval in the computerized interpretation of the cardiotocogram. In conclusion, episodic sampling of the FHR offers advantages in the intrapartum computerized interpre-
580
Wilcox
et al
FHR Variability and Epoch Duration
tation of the cardiotocograph. We suggest an epoch duration of 2 seconds and the adoptation of a single sampled value within this period for measurement of both medium and long-term variability. Because FHR is normally distributed, we suggest it be used instead of R-R intervals. Finally, intrapartum FHR differs from antenatal FHR, and we suggest that this difference be considered when developing computerized systems for interpretation of intrapartum cardiotocograph.
Rt$erences 1 Trimbos JB, Keirse MJNC. Observer variability in assessment of antepartum cardiotocograms. Br J Obstet Gynaecol 1978;85:900-6. 2 Lotgering FK, Wallenburg HCS, Schouten HJA. Interobserver and intraobserver variation in the assessment of antepartum cardiotocograms. Am J Obstet Gynecol 1982;144:701-5. 3 Donker DK, van Geijn HP, Hasman A. Interobserver variation in the assessment of FHR recordings. Eur J Obstet Gynecol Reprod Biol 1993;52:21-8. 4. Pello LC, Rosevear SK, Dawes GS, Moulden M, Redman CWG. Computerized FHR analysis in labor. Obstet Gynecol 1991;78:60210. 5. Mohajer Ml’, Sahota DS, Reed NN, Chang AMZ, Symonds EM, James DK. Cumulative changes in the fetal electrocardiogram and biochemical indices of fetal hypoxia. Eur J Obstet Gynecol Reprod Biol 1994;55:63-70. 6. Siegel S, Castellan NJ. Nonparametric statistics for the behavioural sciences. 2nd ed. New York: McGraw-Hill, 198851-5. 7. Dalton KJ, Dawes GS, Patrick JE. Diurnal, respiratory, and other rhythms of FHR in lambs. Am J Obstet Gynecol 1977;127:414-24. 8. Wheeler T, Cooke E, Mm-rills A. Computer analysis of FHR variation during normal pregnancy. Br J Obstet Gynaecol 1979;86: 186-97. 9. Dawes GS, Moulden M, Redman CWG. Criteria for the design of FHR analysis systems. Int J Biomed Comput 1990;25:287-94. 10. Maeda K. Computerized analysis of cardiotocograms and fetal movements. Baillieres Clin Obstet Gynaecol 1990;4:797-813. 11. Stigsby B, Nielsen PV, Olesen MB, Jaszczak I’, Larsen JF, Lenstrup C. Computer description of the cardiotocogram. 1. The computer program. Acta Obstet Gynecol Stand 1982;109:76-8.
Address
reprint
requests
to:
D. S. Sahota, PhD Department qf Obstetrics and Gynecology Prince qf Wales Hospital Shatin, NT Hong Kong
Receiued August 26, 1996. Received in revisedform December 6, 1996. Accepted Janua y 16, 1997. Copyright 0 1997 by The American College of Obstetricians Gynecologists. Published by Elsevier Science Inc.
and
Obstetrics b Gynecology