- Email: [email protected]
Cesarean delivery rates at the threshold of viability
Cesarean delivery rates at the threshold of viability Mark E. Redman, MD, and Bernard Gonik, MD Detroit, Mich OBJECTIVE: The purpose of this study was to assess whether the treatment of deliveries at the traditional threshold of viability has become more interventional since the publication of higher survival rates. STUDY DESIGN: State vital statistic data from 1995 to 2000 were reviewed to identify deliveries at 22 to 25 weeks of gestation with a birth weight of <1 kg. Demographic data and frequency of cesarean delivery, antepartum bleeding, breech presentation, fetal distress, and cord prolapse were recorded. The cesarean delivery rate and the neonatal mortality rate were calculated for each year. Linear regression for cesarean delivery rate versus year and for neonatal mortality rate versus year were performed at each gestational age. RESULTS: Of 2924 births that met the study criteria, 1005 births (34.4%) were by cesarean delivery. The cesarean delivery rate at 23 weeks rose from 15.9% in 1995 to 28.2% in 2000 (P = .023). Demographic/obstetric factors did not vary with year of delivery. The neonatal mortality rate remained unchanged. CONCLUSION: These data suggest an increasingly aggressive approach to deliveries at the threshold of viability, beginning at 23 weeks of gestation. (Am J Obstet Gynecol 2002;187:873-6.)
Key words: Cesarean delivery rate, viability
In many states, 24 weeks has been defined somewhat arbitrarily as the gestational age at which fetal viability begins. However, investigators have reported recently a range of survival rates below this threshold. In an investigation that was conducted by the Maternal-Fetal Medicine Units Network,1 which evaluated infants with birth weights of <1000 g, neonatal survival rates progressively rose, beginning at gestational ages below the traditional threshold of viability. The results reported by these investigators reveal no obvious threshold effect at 24 weeks of gestation. Furthermore, the authors reported that both survival and intact survival were associated significantly with the obstetrician’s willingness to perform cesarean delivery and opinion whether the fetus was viable. Given these and other data, we hypothesized that current practice patterns would reflect an increase in cesarean delivery of pregnancies at these “nonviable” gestational ages. If other obstetric factors remained unchanged, a changing rate of cesarean delivery over time could be interpreted as evidence of a changing approach to the management of deliveries at the threshold of viability. The objective of this investigation was to assess whether management of deliveries at the traditional From the Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, Wayne State University. Presented at the Twenty-second Annual Meeting of the Society for Maternal-Fetal Medicine, New Orleans, La, January 14-19, 2002. Reprint requests: Mark E. Redman, MD, Department of Obstetrics and Gynecology, Division of Maternal-Fetal Medicine, Hutzel Hospital, 4707 St. Antoine Blvd, Detroit, MI 48201. E-mail: [email protected] © 2002, Mosby, Inc. All rights reserved. 0002-9378/2002 $35.00 + 0 6/6/126985 doi:10.1067/mob.2002.126985
threshold of viability has become more interventional since the publication of higher survival rates, as reflected by a change in cesarean delivery rates. Material and methods Data from the Division of Vital Records and Health Statistics in Michigan (inclusive of the years 1995-2000) were reviewed to identify deliveries at 22 to 25 weeks with birth weights of <1 kg. Gestational age was derived both from last menstrual period and obstetric estimate. Because the latter method correlated most closely with birth weight, it was used for further calculations. Frequencies of vaginal delivery, cesarean delivery, and the following demographic and obstetric factors were recorded for each year: age, racial group, antepartum bleeding, breech presentation, fetal distress, and cord prolapse. The cesarean delivery rate was calculated as 100 cesarean delivery/total births. Stillbirth rate was calculated for deliveries between 22 and 25 weeks of gestation as the number of stillbirths (no heartbeat or respiration at delivery) recorded per 1000 births. Death certificate data were reviewed to ascertain neonatal mortality rate for the study group (neonatal deaths/1000 births) and for the subset of fetuses delivered by cesarean delivery. Association between year of birth and the demographic/obstetric factors listed was assessed by analysis of variance. Linear regression for cesarean delivery rate versus year of birth, for neonatal mortality rate versus year of birth, and for stillbirth rate versus year of birth was performed for each week of gestation. Association between neonatal death and cesarean delivery was assessed by chi-square analysis. Significance was defined as a probability value of <.05. 873
874 Redman and Gonik
October 2002 Am J Obstet Gynecol
Table I. Study population demographic characteristics by year of delivery Characteristic Deliveries (No.) Cesarean deliveries (No.) Cesarean rate (%) Birth weight (%) <500 g 500-999 g Age (%) <20 >40 Black race (%) Bleeding (%) Breech presentation (%) Cord prolapse (%) Fetal distress (%)
1995
1996
1997
1998
1999
2000
Total
P value
488 157 32.2
479 150 31.3
484 168 34.7
460 161 35.0
504 189 37.5
509 180 35.4
2924 1005 34.4
.347 .046 .044
19.5 80.5
22.3 77.7
19.8 80.2
23.5 76.5
21.8 78.2
21.4 78.6
21.4 78.6
.418 .543
18.0 1.6 42.4 8.4 22.1 0.8 8.8
17.5 1.3 42.8 6.3 26.3 2.1 5.6
15.5 2.1 46.3 6.2 26.7 2.3 4.8
18.5 2.8 43.7 6.5 25.7 3.9 7.4
14.7 1.2 43.8 3.6 24.2 1.4 4.6
17.9 2.8 47.2 2.9 20.8 3.3 8.1
17.0 1.9 44.4 5.6 24.2 2.3 6.5
.697 .339 .144 .008 .554 .087 .826
Table II. Cesarean delivery rate* by year of delivery Weeks of gestation 22 (%) 23 (%) 24 (%) 25 (%)
1995
1996
1997
1998
1999
2000
P value
6.9 15.9 42.5 47.9
4.1 15.7 43.8 49.3
7.5 26.0 41.1 51.0
5.6 23.7 45.3 51.1
9.7 25.0 44.8 59.7
4.8 28.2 45.8 50.8
.829 .023 .117 .211
*100 cesarean deliveries/total deliveries.
Table III. Neonatal mortality rate* by year of delivery Weeks of gestation 22 23 24 25
1995
1996
1997
1998
1999
2000
P value
919.5 690.3 462.7 233.1
948.5 675.9 354.2 186.6
925.0 707.3 425.5 206.9
887.6 677.4 395.7 266.2
903.2 726.6 324.1 230.2
923.1 663.6 516.9 231.6
.432 .961 .830 .481
*Neonatal deaths/1000 live births.
Zone charts were also constructed for cesarean delivery rate versus year of delivery at each week of gestation studied. Zone charts, adapted from the discipline of “statistical process control,” graphically demonstrate whether there is change in the level of a process over time. The center horizontal line of a zone chart demonstrates the mean value or reference value of a process under evaluation; the other horizontal lines represent 1, 2, or 3 SDs (σ) above or below the reference value. The probability that a point will be found at >3 σ because of chance alone is approximately 0.00135, and 3 σ customarily represents significant deviation from the reference value in process control charts. Each observation is assigned a zone score of 0 if the observation is between the mean and 1 σ, of 2 if the observation is between 1 σ and 2 σ, of 4 if the observation is between 2 σ and 3 σ, and of 8 if the observation is >3 σ. The scores are cumulative within a given chart and are found within the points on the chart.2
Results From 1995 through 2000, 796,238 deliveries were recorded in Michigan, of which 2924 (0.367%) occurred between 22 and 25 weeks of gestation. Among the deliveries between 22 and 25 weeks of gestation, 1005 deliveries (34.4%) were cesarean deliveries. The demographic characteristics of the study population in each year are described in Table I. None of the demographic factors varied with year of delivery. None of the obstetric factors that are associated commonly with an increased likelihood of cesarean delivery increased during the study period. Bleeding, the sole maternal indication for cesarean delivery that was studied, demonstrated a small but statistically significant decrease during the study period. The cesarean delivery rate at 23 weeks of gestation significantly increased from 15.9% in 1995 to 28.2% in 2000 (P = .023). Table II describes the cesarean delivery rate by year of delivery and the significance of the association between year of delivery and cesarean delivery rate for each
Redman and Gonik 875
Volume 187, Number 4 Am J Obstet Gynecol
Table IV. Neonatal mortality rate* by delivery method Weeks of gestation 22 23 24 25
Vaginal
Cesarean
P value
935.9 736.1 438.2 229.2
657.1 539.5 372.2 222.9
<.001 <.001 .060 .789
*Neonatal deaths/1000 live births.
week of gestation. The Figure contains a zone chart that depicts the cesarean delivery rate at 23 weeks of gestation for each year compared with the rate and SD from 1995. The chart illustrates a clear shift in cesarean delivery rates beginning in 1997. The zone score of 8 for the third observation, representing the 1997 cesarean delivery rate, indicates the beginning of a new “process.” The zone charts for the rates of cesarean delivery at 24 and 25 weeks of gestation appear to demonstrate shifts in 1998 and 1997, respectively, but the changes are not statistically significant (not shown), and there is no trend at 22 weeks of gestation (not shown). The overall neonatal mortality rate for the study group was 523.7 per 1000 live births and 279.2 per 1000 live births by cesarean delivery. The neonatal mortality rate for each year is tabulated by week of gestation in Table III. There was no significant association between year of delivery and neonatal mortality rate at any of the gestational ages that were assessed. The neonatal mortality rates at each week of gestation that were studied are tabulated by delivery method in Table IV, which reveals lower mortality rates in those fetuses delivered by cesarean delivery at 22 and 23 weeks of gestation. The stillbirth rate at each week of gestation did not change during the study period (Table V). Comment The cesarean delivery rate at 23 weeks significantly increased, although the rates at other gestational ages did not change from 1995 to 2000. Demographic and obstetric factors did not change during the study period and were not associated with cesarean delivery rate, which suggests that deliveries at the threshold of viability are being treated in a more interventional fashion, beginning at 23 weeks of gestation. The rise in cesarean delivery rates was not associated with a change in fetal or neonatal mortality rates over time. Other investigators reported higher rates of survival and intact survival after delivery around the threshold of viability during the 1990s, compared with data based on previous studies.1,3-5 Bottoms et al1 reported data that showed not only that survival gradually increased through a range of gestational ages but also that a physician’s view of the pregnancy as viable and the physician’s willingness to perform cesarean delivery significantly in-
Figure. Zone chart of cesarean delivery rate at 23 weeks of gestation versus the year of delivery. Observations 1, 2, 3, 4, 5, and 6 represent the 23-week cesarean delivery rates for 1995, 1996, 1997, 1998, 1999, and 2000, respectively. The circle represents each observation and is labeled with its respective zone score.
creased the likelihood of both survival and intact survival. Because the study did not find an association between actual cesarean delivery and survival, the authors concluded that the approach to obstetric management (ie, treating the patient as if the fetus were viable) significantly increased survival and survival without major morbidity. The publication of this Maternal-Fetal Medicine Units Network study in 1997 appears to coincide with the beginning of the rise in cesarean delivery rate at 23 weeks of gestation that we found in Michigan. Our data appear to show that management of 23-week deliveries in Michigan has grown increasingly aggressive since 1997 because cesarean delivery rates rose significantly although other obstetric and demographic factors remained unchanged. These data may reflect an evolving perception of the gestational age limits to fetal viability. An analysis of potential predictors of obstetric treatment at the threshold of viability (such as institution, academic affiliation, and level of nursery care) could provide further insight into the causes of the significant increase in cesarean delivery rates at 23 weeks of gestation. The rise in cesarean delivery rates at 23 weeks of gestation was not accompanied by a change in neonatal mortality rates over time during the study period. The neonatal mortality rate may have remained constant for several reasons. The method of delivery may not influence neonatal mortality rate for fetuses near the threshold of viability, or there may have been a change in neonatal mortality rates for the entire study group or for subgroups that this investigation did not detect. One may theorize that increasing cesarean delivery rates at 23 weeks of gestation might “substitute” fetal death with neonatal death by preventing intrapartum demise for fetuses who will not survive after delivery. Our data, however, do not demonstrate this phenomenon because the
876 Redman and Gonik
October 2002 Am J Obstet Gynecol
Table V. Stillbirth rate* by year of delivery Weeks of gestation 22 23 24 25
1995
1996
1997
1998
1999
2000
P value
463.0 319.6 198.2 109.2
419.2 350.6 207.1 128.3
503.1 349.4 172.4 88.4
473.4 267.9 185.0 115.9
465.5 354.2 210.2 136.6
455.5 260.1 161.8 91.5
.790 .361 .383 .851
*Stillbirths/1000 births.
stillbirth rates presented in Table V remained unchanged during the study period. Whether cesarean delivery itself or the willingness to intervene for fetal indications directly affected neonatal mortality rates is not established by the data we present. Similarly, we are unable to discern from our data whether cesarean delivery near the threshold of viability improved the outcome for fetuses in the study population. Although the lower neonatal mortality rates in those fetuses delivered by cesarean delivery at 22 and 23 weeks of gestation (Table IV) are suggestive, conclusions about causation cannot be drawn because of an inability to control for selection bias and other confounding variables in this retrospective study. Because neonatal death was not associated with method of delivery at 24 and 25 weeks of gestation, one interpretation is that the difference that was observed at 22 and 23 weeks of gestation reflects bias in the decisions that were made because of factors that were not evaluated in this investigation. Increases in maternal morbidity and in financial costs may be associated with rising cesarean delivery rates at the threshold of viability, but our study does not provide a basis for evaluating these trends. Additionally, our data do not provide a basis for the comparison of morbidity rates by delivery route among these patients. These issues may be subjected to future investigation, now that the trend of increasing cesarean delivery at the threshold of viability has been presented. The gestational age limit for fetal viability has evolved over time along with advancements in obstetric and neonatal treatment, such as the antenatal administration
of corticosteroids and the neonatal use of exogenous surfactant. The data from this investigation illustrate that a more interventional approach to delivery before the traditional threshold of viability has risen in frequency coincident with data that describe progressively increasing survival rates before and after 24 weeks of gestation. The impact that increasingly aggressive obstetric management of 23-week deliveries will have on the definition of fetal viability beginning at 24 weeks of gestation remains undetermined. We thank Mr. Glenn Copeland and his associates for their substantial contribution in data compilation and statistical analysis.
REFERENCES
1. Bottoms SF, Paul RH, Iams JD, Mercer BM, Thom EA, Roberts JM, et al. Obstetric determinants of neonatal survival: influence of willingness to perform cesarean delivery on survival of extremely low-birth-weight infants. Am J Obstet Gynecol 1997; 176:950-6. 2. Montgomery DC. Introduction to statistical quality control. 2nd ed. New York: John Wiley; 1991. 3. El-Metwally D, Vohr B, Tucker R. Survival and neonatal morbidity at the limits of viability in the mid 1990s: 22 to 25 weeks. J Pediatr 2000;137:616-22. 4. Costeloe K, Hennessy E, Gibson AT, Marlow N, Wilkinson AR, for the EPICure Study Group. The EPICure study: outcomes to discharge from hospital for infants born at the threshold of viability. Pediatrics 2000;106:659-71. 5. Lemons JA, Bauer CR, Korones SB, Papile LA, Stoll BJ, Verter J, et al. Very low birth weight outcomes for the National Institute of Child Health and Human Development Neonatal Research Network. Pediatrics 2001;107:E1.
Key words: Cesarean delivery rate, viability
In many states, 24 weeks has been defined somewhat arbitrarily as the gestational age at which fetal viability begins. However, investigators have reported recently a range of survival rates below this threshold. In an investigation that was conducted by the Maternal-Fetal Medicine Units Network,1 which evaluated infants with birth weights of <1000 g, neonatal survival rates progressively rose, beginning at gestational ages below the traditional threshold of viability. The results reported by these investigators reveal no obvious threshold effect at 24 weeks of gestation. Furthermore, the authors reported that both survival and intact survival were associated significantly with the obstetrician’s willingness to perform cesarean delivery and opinion whether the fetus was viable. Given these and other data, we hypothesized that current practice patterns would reflect an increase in cesarean delivery of pregnancies at these “nonviable” gestational ages. If other obstetric factors remained unchanged, a changing rate of cesarean delivery over time could be interpreted as evidence of a changing approach to the management of deliveries at the threshold of viability. The objective of this investigation was to assess whether management of deliveries at the traditional From the Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, Wayne State University. Presented at the Twenty-second Annual Meeting of the Society for Maternal-Fetal Medicine, New Orleans, La, January 14-19, 2002. Reprint requests: Mark E. Redman, MD, Department of Obstetrics and Gynecology, Division of Maternal-Fetal Medicine, Hutzel Hospital, 4707 St. Antoine Blvd, Detroit, MI 48201. E-mail: [email protected] © 2002, Mosby, Inc. All rights reserved. 0002-9378/2002 $35.00 + 0 6/6/126985 doi:10.1067/mob.2002.126985
threshold of viability has become more interventional since the publication of higher survival rates, as reflected by a change in cesarean delivery rates. Material and methods Data from the Division of Vital Records and Health Statistics in Michigan (inclusive of the years 1995-2000) were reviewed to identify deliveries at 22 to 25 weeks with birth weights of <1 kg. Gestational age was derived both from last menstrual period and obstetric estimate. Because the latter method correlated most closely with birth weight, it was used for further calculations. Frequencies of vaginal delivery, cesarean delivery, and the following demographic and obstetric factors were recorded for each year: age, racial group, antepartum bleeding, breech presentation, fetal distress, and cord prolapse. The cesarean delivery rate was calculated as 100 cesarean delivery/total births. Stillbirth rate was calculated for deliveries between 22 and 25 weeks of gestation as the number of stillbirths (no heartbeat or respiration at delivery) recorded per 1000 births. Death certificate data were reviewed to ascertain neonatal mortality rate for the study group (neonatal deaths/1000 births) and for the subset of fetuses delivered by cesarean delivery. Association between year of birth and the demographic/obstetric factors listed was assessed by analysis of variance. Linear regression for cesarean delivery rate versus year of birth, for neonatal mortality rate versus year of birth, and for stillbirth rate versus year of birth was performed for each week of gestation. Association between neonatal death and cesarean delivery was assessed by chi-square analysis. Significance was defined as a probability value of <.05. 873
874 Redman and Gonik
October 2002 Am J Obstet Gynecol
Table I. Study population demographic characteristics by year of delivery Characteristic Deliveries (No.) Cesarean deliveries (No.) Cesarean rate (%) Birth weight (%) <500 g 500-999 g Age (%) <20 >40 Black race (%) Bleeding (%) Breech presentation (%) Cord prolapse (%) Fetal distress (%)
1995
1996
1997
1998
1999
2000
Total
P value
488 157 32.2
479 150 31.3
484 168 34.7
460 161 35.0
504 189 37.5
509 180 35.4
2924 1005 34.4
.347 .046 .044
19.5 80.5
22.3 77.7
19.8 80.2
23.5 76.5
21.8 78.2
21.4 78.6
21.4 78.6
.418 .543
18.0 1.6 42.4 8.4 22.1 0.8 8.8
17.5 1.3 42.8 6.3 26.3 2.1 5.6
15.5 2.1 46.3 6.2 26.7 2.3 4.8
18.5 2.8 43.7 6.5 25.7 3.9 7.4
14.7 1.2 43.8 3.6 24.2 1.4 4.6
17.9 2.8 47.2 2.9 20.8 3.3 8.1
17.0 1.9 44.4 5.6 24.2 2.3 6.5
.697 .339 .144 .008 .554 .087 .826
Table II. Cesarean delivery rate* by year of delivery Weeks of gestation 22 (%) 23 (%) 24 (%) 25 (%)
1995
1996
1997
1998
1999
2000
P value
6.9 15.9 42.5 47.9
4.1 15.7 43.8 49.3
7.5 26.0 41.1 51.0
5.6 23.7 45.3 51.1
9.7 25.0 44.8 59.7
4.8 28.2 45.8 50.8
.829 .023 .117 .211
*100 cesarean deliveries/total deliveries.
Table III. Neonatal mortality rate* by year of delivery Weeks of gestation 22 23 24 25
1995
1996
1997
1998
1999
2000
P value
919.5 690.3 462.7 233.1
948.5 675.9 354.2 186.6
925.0 707.3 425.5 206.9
887.6 677.4 395.7 266.2
903.2 726.6 324.1 230.2
923.1 663.6 516.9 231.6
.432 .961 .830 .481
*Neonatal deaths/1000 live births.
Zone charts were also constructed for cesarean delivery rate versus year of delivery at each week of gestation studied. Zone charts, adapted from the discipline of “statistical process control,” graphically demonstrate whether there is change in the level of a process over time. The center horizontal line of a zone chart demonstrates the mean value or reference value of a process under evaluation; the other horizontal lines represent 1, 2, or 3 SDs (σ) above or below the reference value. The probability that a point will be found at >3 σ because of chance alone is approximately 0.00135, and 3 σ customarily represents significant deviation from the reference value in process control charts. Each observation is assigned a zone score of 0 if the observation is between the mean and 1 σ, of 2 if the observation is between 1 σ and 2 σ, of 4 if the observation is between 2 σ and 3 σ, and of 8 if the observation is >3 σ. The scores are cumulative within a given chart and are found within the points on the chart.2
Results From 1995 through 2000, 796,238 deliveries were recorded in Michigan, of which 2924 (0.367%) occurred between 22 and 25 weeks of gestation. Among the deliveries between 22 and 25 weeks of gestation, 1005 deliveries (34.4%) were cesarean deliveries. The demographic characteristics of the study population in each year are described in Table I. None of the demographic factors varied with year of delivery. None of the obstetric factors that are associated commonly with an increased likelihood of cesarean delivery increased during the study period. Bleeding, the sole maternal indication for cesarean delivery that was studied, demonstrated a small but statistically significant decrease during the study period. The cesarean delivery rate at 23 weeks of gestation significantly increased from 15.9% in 1995 to 28.2% in 2000 (P = .023). Table II describes the cesarean delivery rate by year of delivery and the significance of the association between year of delivery and cesarean delivery rate for each
Redman and Gonik 875
Volume 187, Number 4 Am J Obstet Gynecol
Table IV. Neonatal mortality rate* by delivery method Weeks of gestation 22 23 24 25
Vaginal
Cesarean
P value
935.9 736.1 438.2 229.2
657.1 539.5 372.2 222.9
<.001 <.001 .060 .789
*Neonatal deaths/1000 live births.
week of gestation. The Figure contains a zone chart that depicts the cesarean delivery rate at 23 weeks of gestation for each year compared with the rate and SD from 1995. The chart illustrates a clear shift in cesarean delivery rates beginning in 1997. The zone score of 8 for the third observation, representing the 1997 cesarean delivery rate, indicates the beginning of a new “process.” The zone charts for the rates of cesarean delivery at 24 and 25 weeks of gestation appear to demonstrate shifts in 1998 and 1997, respectively, but the changes are not statistically significant (not shown), and there is no trend at 22 weeks of gestation (not shown). The overall neonatal mortality rate for the study group was 523.7 per 1000 live births and 279.2 per 1000 live births by cesarean delivery. The neonatal mortality rate for each year is tabulated by week of gestation in Table III. There was no significant association between year of delivery and neonatal mortality rate at any of the gestational ages that were assessed. The neonatal mortality rates at each week of gestation that were studied are tabulated by delivery method in Table IV, which reveals lower mortality rates in those fetuses delivered by cesarean delivery at 22 and 23 weeks of gestation. The stillbirth rate at each week of gestation did not change during the study period (Table V). Comment The cesarean delivery rate at 23 weeks significantly increased, although the rates at other gestational ages did not change from 1995 to 2000. Demographic and obstetric factors did not change during the study period and were not associated with cesarean delivery rate, which suggests that deliveries at the threshold of viability are being treated in a more interventional fashion, beginning at 23 weeks of gestation. The rise in cesarean delivery rates was not associated with a change in fetal or neonatal mortality rates over time. Other investigators reported higher rates of survival and intact survival after delivery around the threshold of viability during the 1990s, compared with data based on previous studies.1,3-5 Bottoms et al1 reported data that showed not only that survival gradually increased through a range of gestational ages but also that a physician’s view of the pregnancy as viable and the physician’s willingness to perform cesarean delivery significantly in-
Figure. Zone chart of cesarean delivery rate at 23 weeks of gestation versus the year of delivery. Observations 1, 2, 3, 4, 5, and 6 represent the 23-week cesarean delivery rates for 1995, 1996, 1997, 1998, 1999, and 2000, respectively. The circle represents each observation and is labeled with its respective zone score.
creased the likelihood of both survival and intact survival. Because the study did not find an association between actual cesarean delivery and survival, the authors concluded that the approach to obstetric management (ie, treating the patient as if the fetus were viable) significantly increased survival and survival without major morbidity. The publication of this Maternal-Fetal Medicine Units Network study in 1997 appears to coincide with the beginning of the rise in cesarean delivery rate at 23 weeks of gestation that we found in Michigan. Our data appear to show that management of 23-week deliveries in Michigan has grown increasingly aggressive since 1997 because cesarean delivery rates rose significantly although other obstetric and demographic factors remained unchanged. These data may reflect an evolving perception of the gestational age limits to fetal viability. An analysis of potential predictors of obstetric treatment at the threshold of viability (such as institution, academic affiliation, and level of nursery care) could provide further insight into the causes of the significant increase in cesarean delivery rates at 23 weeks of gestation. The rise in cesarean delivery rates at 23 weeks of gestation was not accompanied by a change in neonatal mortality rates over time during the study period. The neonatal mortality rate may have remained constant for several reasons. The method of delivery may not influence neonatal mortality rate for fetuses near the threshold of viability, or there may have been a change in neonatal mortality rates for the entire study group or for subgroups that this investigation did not detect. One may theorize that increasing cesarean delivery rates at 23 weeks of gestation might “substitute” fetal death with neonatal death by preventing intrapartum demise for fetuses who will not survive after delivery. Our data, however, do not demonstrate this phenomenon because the
876 Redman and Gonik
October 2002 Am J Obstet Gynecol
Table V. Stillbirth rate* by year of delivery Weeks of gestation 22 23 24 25
1995
1996
1997
1998
1999
2000
P value
463.0 319.6 198.2 109.2
419.2 350.6 207.1 128.3
503.1 349.4 172.4 88.4
473.4 267.9 185.0 115.9
465.5 354.2 210.2 136.6
455.5 260.1 161.8 91.5
.790 .361 .383 .851
*Stillbirths/1000 births.
stillbirth rates presented in Table V remained unchanged during the study period. Whether cesarean delivery itself or the willingness to intervene for fetal indications directly affected neonatal mortality rates is not established by the data we present. Similarly, we are unable to discern from our data whether cesarean delivery near the threshold of viability improved the outcome for fetuses in the study population. Although the lower neonatal mortality rates in those fetuses delivered by cesarean delivery at 22 and 23 weeks of gestation (Table IV) are suggestive, conclusions about causation cannot be drawn because of an inability to control for selection bias and other confounding variables in this retrospective study. Because neonatal death was not associated with method of delivery at 24 and 25 weeks of gestation, one interpretation is that the difference that was observed at 22 and 23 weeks of gestation reflects bias in the decisions that were made because of factors that were not evaluated in this investigation. Increases in maternal morbidity and in financial costs may be associated with rising cesarean delivery rates at the threshold of viability, but our study does not provide a basis for evaluating these trends. Additionally, our data do not provide a basis for the comparison of morbidity rates by delivery route among these patients. These issues may be subjected to future investigation, now that the trend of increasing cesarean delivery at the threshold of viability has been presented. The gestational age limit for fetal viability has evolved over time along with advancements in obstetric and neonatal treatment, such as the antenatal administration
of corticosteroids and the neonatal use of exogenous surfactant. The data from this investigation illustrate that a more interventional approach to delivery before the traditional threshold of viability has risen in frequency coincident with data that describe progressively increasing survival rates before and after 24 weeks of gestation. The impact that increasingly aggressive obstetric management of 23-week deliveries will have on the definition of fetal viability beginning at 24 weeks of gestation remains undetermined. We thank Mr. Glenn Copeland and his associates for their substantial contribution in data compilation and statistical analysis.
REFERENCES
1. Bottoms SF, Paul RH, Iams JD, Mercer BM, Thom EA, Roberts JM, et al. Obstetric determinants of neonatal survival: influence of willingness to perform cesarean delivery on survival of extremely low-birth-weight infants. Am J Obstet Gynecol 1997; 176:950-6. 2. Montgomery DC. Introduction to statistical quality control. 2nd ed. New York: John Wiley; 1991. 3. El-Metwally D, Vohr B, Tucker R. Survival and neonatal morbidity at the limits of viability in the mid 1990s: 22 to 25 weeks. J Pediatr 2000;137:616-22. 4. Costeloe K, Hennessy E, Gibson AT, Marlow N, Wilkinson AR, for the EPICure Study Group. The EPICure study: outcomes to discharge from hospital for infants born at the threshold of viability. Pediatrics 2000;106:659-71. 5. Lemons JA, Bauer CR, Korones SB, Papile LA, Stoll BJ, Verter J, et al. Very low birth weight outcomes for the National Institute of Child Health and Human Development Neonatal Research Network. Pediatrics 2001;107:E1.