Is it time to reassess the risk for the growth-retarded fetus with normal Doppler velocimetry of the umbilical artery?

Is it time to reassess the risk for the growth-retarded fetus with normal Doppler velocimetry of the umbilical artery? Emanuel P. Gaziano, MD, Heather Knox, BA, Bruce Ferrera, MD, Debra G. Brandt, RDMS, Steven E. Calvin, MD, and G. Eric Knox, MD Minneapolis, Minnesota OBJECTIVE: Previous studies demonstrate an association between abnormal umbilical artery Doppler velocimetry and the birth of a small-for-gestational-age infant and between abnormal result and adverse neonatal outcome. The hypothesis is that preterm growth-retarded infants with normal antenatal velocimetry have outcomes similar to other preterm infants, whereas preterm small-for-gestational-age infants with abnormal Doppler results define a subgroup with increased morbidity. STUDY DESIGN: For 100 live-born infants, at risk for fetal growth retardation and undergoing antenatal Doppler and targeted ultrasonographic examinations, we assessed a number of complete neonatal outcome parameters. RESULTS: Ten neonatal deaths occurred in the study population, seven with abnormal Doppler results and three with normal Doppler results. Of the 90 surviving infants, gestational age at delivery was not different between the Doppler normal and abnormal neonates, whereas birth weight (1714 gm vs 1379 gm) was higher in the Doppler normal group (p = 0.006). The presence of intraventricular hemorrhage (20% vs 6%) was higher in the abnormal group (p = 0.05). Abnormal Doppler results defined an infant group destined for prolonged hospitalization, mean intensive care days (21 vs 9), and special care nursery days (25 vs 9). Thirty-eight percent of small-for-gestational-age babies had a normal Doppler result. Analysis of variance indicated small-for-gestational-age infants with abnormal Doppler results (n = 20) had a mean intensive care unit stay of 31 days, significantly different (p = 0.005) from small-for-gestational-age infants with normal Doppler results (n = 14), non-small-for-gestational-age infants with abnormal results (n = 21), and non-small-for-gestational-age infants with normal results (n = 35) whose mean intensive care unit stays were 14, 12, and 7 days, respectively. Gestational age at delivery (33.0 weeks) was not different among these groupings, not accounting for the observed differences. CONCLUSION: Normal antenatal velocimetry defines a distinct subgroup of preterm small-for-gestational-age infants at less risk for prolonged hospitalization compared with those with abnormal velocimetry. (AM J OBSTET GYNECOL 1994;170:1734-43.)

Key words: Fetal growth retardation Doppler umbilical artery, neonatal outcome

Abnormal umbilical artery velocimetry is associated with the birth of small-for-gestational-age (SeA) infants and those with certain fetal anomalies and chromosome defects.'·4 Although the role for clinical fetal Doppler studies continues to be explored, studies suggest ultrasonographic biometry to be superior to umbilical artery Doppler for antenatal seA diagnosis, and reviews of recent investigations suggest a likely association between abnormal fetal Doppler velocimetry and increased neonatal mortality and morbidity.s Many fetal Doppler outcome studies are concerned From the Perinatology and Neonatology Service, Abbott-Northwestern Hospital and Minneapolis Children's Medical Center. Presented at the Sixty-first Annual Meeting of The Central Association of Obstetricians and Gynecologists, White Sulphur Springs, West Virginia, October 28-30, 1993. Reprint requests: Emanuel Gaziano, MD, The Perinatal Center, Abbott-Northwestern Hospital, 800 E. 28th St., Minneapolis, MN 55407. Copyright © 1994 by Mosby-Year Book, Inc. 0002-9378/$3.00 + 0 6/6/55078

1734

with SeA prediction and define morbidity broadly and nonspecifically, confining index values to neonatal intensive care unit admissions, Apgar score, birth weight percentile, umbilical cord gases, fetal distress, cesarean sections, and the presence of clinical factors such as hypertension. 6 .• Investigators note a variety of morbidity issues, including the degree of Doppler abnormality and adverse neonatal events. 9 Extremely poor outcome is reported with zero or reversal of flow in diastole. '0 Burke et al. 11 indicate that growth retardation in the presence of normal umbilical artery Doppler may be a benign condition, but the outcome variables measured were perinatal deaths, fetal distress, and pre term deliveries and not neonatal outcomes in surviving infants. Although these associations are well documented, further refinement is needed to assess neonatal risks for the seA infant on the basis of antenatal Doppler velocimetry. The purpose of this study is to determine the incidence of normal umbilical artery velocimetry in the

Gaziano et aL

Volume 170, Number 6 Am J Obstet Gynecol

1735

Table I. Clinical characteristics of neonatal deaths occurring in patients suspected of intrauterine growth retardation (lUGR) undergoing antenatal Doppler studies of the umbilical artery and targeted ultrasonographic examinations (n = 10) Gestational age (wk)

SID ratio

27

6.7

900

2 3 4

31 26 28

4.0 5.1 Reverse flow

5 6

26 32

7

Case No.

Birth weight (gm)

RDS

SGA

Comments

Severe

No

1150 760 480

Severe Severe Severe

Yes No Yes

Pulmonary hypertension, tricuspid regurgitation Cytomegalovirus

5.1 5.2

760 1240

Severe Moderate

No No

26

4.2

430

Severe

Yes

8

26

3.0

465

Severe

Yes

9

25

3.2

740

Moderate

No

10

26

3.8

1000

Moderate

No

Small ventricular septal defect Early neonatal death Hydrops, twin, normal chromosomes Abnormal head ultrasonography Severe hemolytic disease Abnormal head ultrasonography Hydrops, twin

Cases 1 through 7 represent abnormal SID values, cases 8 to 10 normal values.

pre term SGA neonate and to examine its relationship to groupings of patients on the basis of antenatal Doppler result and neonatal birth weight status, correlating these groupings with a number of neonatal outcome parameters. Assessing neonatal outcome is important, because fetal Doppler data may suggest which at-risk fetuses may be modified as candidates for intensive surveillance. Material and methods

Over a 47-month period targeted ultrasonographic and fetal Doppler measurements were performed on high-risk pregnancies referred to the Abbott-Northwestern/Minneapolis Children's Perinatal Center, a regional tertiary site. During this time 100 consecutive live-born infants were selected for study. All were clinically suspected to be SGA on the basis of the results of targeted ultrasonographic examinations, which included an echocardiogram and detailed biometry. In addition, pulsed Doppler recordings of the umbilical artery were obtained. Criteria for suspected SGA were determined by the physician performing the scan and included the presence of ultrasonographically estimated fetal weight ::; 10th percentile for gestational age. Only neonates with complete ultrasonographic, umbilical artery Doppler, and neonatal follow-up data were included. More than 200 Doppler or ultrasonographic examinations were made in the study population, whereas only the last Doppler measurement taken before delivery was considered for the analysis. Included in the consecutive data set are fetuses from twin gestations who were clinically suspected to be growth

retarded, 27 in the Doppler normal group and 18 in the Doppler abnormal group. During this study period the Doppler result was known to the clinician, but in no instance was a decision for delivery made solely on the basis of the results of Doppler values. Mean examination-to-delivery interval was 5.2 days, and in every case but one delivery occurred within 18 days of testing. No infants with lethal anomalies were studied, although some neonates had a number of associated findings such as partial urinary obstruction, hypospadias, hydrocele, and imperforate anus. Similar findings were present in both Doppler and SGA groups, and their distribution was not significantly different among the different groupings. One case of trisomy 21 was noted in the Doppler abnormal group. Pulsed Doppler recordings were obtained from a mid segment of fetal umbilical artery during quiet fetal state with either an ATL Ultramark 8 (Advanced Technology Laboratory, Bothell, Wash.) or an Acuson 128 X-P/lO (Acuson, Mountain View, Calif.) with either a 3 or 5 MHz transducer and a sample volume size of 1.5 mm 2 • Three to five reproducible waveforms were analyzed, and a systolic/diastolic (SID) ratio was calculated by measuring peak SID values by calipers. The Doppler value was considered to be abnormal if the mean SID ratio was ~ 2 SDs above the mean for gestational age. These SID values were derived from a gestational age-specific nomogram constructed from 274 normal singleton pregnancies from our own population. 12 Abnormal Doppler results included those with zero or reversal of flow in diastole. Of the 100 patients, 42 were considered to have normal antenatal pulsed Doppler values and 48

1736 Gaziano et al.

June 1994

Am J Obstet Gyneco1

% Requiring Delivery Room Resuscitation

0/. with RDS on Initial Chest X-ray

J~~I2I2~~.I.1]

NS

+!.:::.:::.::.-_...::.;,_==:::u----,

NS

~~------------~ % with Abnormal

Head Ultrasound

% Requiring

IVTherapy

+""...a__--,

P.0.05

•••

~~~=== ~.I.I _ _ _ _--, P
I

60%

40%

20%

0%

~

CJ Doppler Normal

0

80%

tOO

Doppler Abnormal

Fig. 1. Outcome characteristics of survIVIng neonates suspected of having IUGR on targeted ultrasonography grouped according to antenatal umbilical artery Doppler values(N = 90). RDS, Respiratory distress syndrome; IV, intravenous. Asterisk, p :5 0.05, statistically significant.

I

8.7 NICU

'.

21

I

I

8.6 SCU

J

25.2

o

10

5

PeG.OS

20

15

25

PeG.OS

30

Duration of Hosp~alizatlon (Days)

o

UADNormai

o

UAD Abnormal

Fig. 2. Neonatal intensive care unit (NICU) and special care unit (SCU) length of stay (days) for neonates suspected of having IUGR on the basis of antenatal Doppler and ultrasonography studies (n = 90). UAD, Umbilical artery Doppler SID ratio. Asterisk, p < 0.05, statistically significant.

were abnormal, seven of whom demonstrated no or reversal of flow in diastole. Detailed neonatal data were also compiled and included the following: stillbirths, neonatal deaths, and presence and type of anomaly. For survivors the following data were obtained: gestational' age, birth weight, Apgar scores, presence of SGA infant, need for delivery room resuscitation, presence of respiratory distress syndrome (RDS), cerebral ultrasonography results, duration of mechanical ventilation, duration of intravenous therapy, duration of oxygen therapy, and length of stay in the neonatal intensive care unit. An intermediate step-down neonatal care unit, termed special care unit, was used for neonates who were premature and needed supportive care but not mechanical ventilation. SGA status is defined as birth weight SlOth percen-

tile for gestational age. 12 Resuscitation was deemed present when intubation or mask bagging occurred or when any other forms of initial neonatal support were required other than that of simple bulb suctioning. RDS was considered to be present if the initial chest x-ray film demonstrated typical findings for this condition. Head ultrasonographic examinations were considered to be abnormal in the presence of evidence for any grade of intraventricular hemorrhage. The data were tabulated and analyzed according to relationships between normal and abnormal pulsed Doppler results and neonatal outcome parameters. Means were calculated for the variables or groups of variables considered. Where appropriate, t tests, medians, standard deviations, and X2 values were assessed and compared. Analysis of variance and Duncan's multiple range tests were performed for variables that were tested for significance among various Doppler result and birth weight status groupings. Results

Of the 100 fetuses suspected clinically to be growth retarded, 38 were SGA with a birth weight slOth percentile for gestational age. Ten neonatal deaths occurred in the study population. Of the 90 surviving infants 14 were SGA with normal antenatal Doppler results, 20 were SGA with abnormal Doppler results, 21 were nO!1-SGA with abnormal Doppler results, and 35 were non-SGA with normal Doppler results. Thirty-nine percent of the SGA infants had a normal antenatal Doppler result. Table I shows the clinical characteristics of the 10 neonatal deaths, seven occurring in the Doppler abnormal group. The only neonatal death in an SGA infant with a normal Doppler result occurred in a neonate

Gaziano et al.

Volume 170, Number 6 Am J Obstet Gynecol

1737

Table II. Gestational age and birth weight of surviving neonates suspected of IUGR on targeted ultrasonography, grouped according to antenatal umbilical artery Doppler values (N = 90) Doppler values Normal (n

= 49)

I

Abnormal (n SD

Mean

33.3

2.9

1714

570

Variables

Mean

Gestational age at delivery (wk) Birth weight (gm)

= 41)

I

SD

Significance

32.5

3.7

NS

1379

548

p < 0.01*

NS, Not significant.

*Highly statistically significant.

Table III. Characteristics of infants born SGA grouped according to antenatal umbilical artery values Doppler result Variables

Gestational age at delivery (wk) Test-to-delivery interval (days) Birth weight (gm) Neonatal hypoglycemia (%) Abnonnal head ultrasonography (%)* Admission to neonatal intensive care unit (%) Days (mean) convalescence in special care nursery

Normal (n = 14)

I

Abnormal (n = 20)

Significance

33.6

32.4

NS

7.3

4.0

NS

1414 57

1157 85

o

5

43

80

p<

38.2

p < 0.05t

5.5

NS

P = 0.07 NS

0.05t

*Any degree of intraventricular hemorrhage. tStatistically significant.

'with severe hemolytic disease. Six of seven neonates with abnormal Doppler results and severe RDS had early neonatal deaths. Of the 90 surviving infants suspected to have growth retardation, 49 had a normal antenatal Doppler result and 41 were abnormal. The gestational age at delivery (32.9 weeks) was not different between the two Doppler groups, but, as expected, neonates with abnormal Doppler results had a significantly lower birth weight (1379 gm vs 1714 gm,p < 0.01) (Table II). No differences were noted in Apgar scores at 1 and 5 minutes for surviving neonates. Fig. 1 shows no significant correlations between Doppler abnormal and normal results and the number of infants requiring delivery room resuscitation (54% vs 53%) or the presence of RDS on initial chest x-ray film (51% vs 41%). The number of patients with any degree of intraventricular hemorrhage was higher in the Doppler abnormal group (19.5% vs 6%, P = 0.05), whereas the incidence of infants requiring any form of intravenous therapy was higher in the Doppler abnormal group (85% vs

63%,P < 0.05). Although the Doppler abnormal group overall showed a trend toward increased mean duration of mechanical ventilation (24 vs 12 days) and oxygen therapy (34 vs 21 days), these differences did not reach statistical significance. Neonates suspected of being SGA who had abnormal Doppler findings compared with SGA infants with normal Doppler results had a significantly prolonged hospitalization (neonatal intensive care unit 21 vs 8.7 days, p < 0.05, and special care unit 25.2 vs 8.6 days, p < 0.05) (Fig. 2). Fig. 3 graphically displays the prolonged length of stay grouped according to umbilical artery values. Twice as many (37% vs 18%) of the SGA neonates had ~ 15 days in the neonatal intensive care unit when the antenatal Doppler value was abnormal compared with when it was normal. Table III shows that of the 34 surviving neonates who were born SGA, mean gestational age at delivery (33.0 weeks) and mean test-to-delivery intervals (5.6 days) were not different between the Doppler normal and

1738

Gaziano et al.

June 1994 Am J Obstet Gynecol

70~--------------------------------' (61)

60 50 Neonates Treated in NICU(%)

40

,~,--

(39) _- ..---___

-------.(37)

30

P
20

(20)

(18)

10+----,--~-r----~----._----r_--~ 1-14 >=15 o Duration of NICU Hospitalization (Days) ~

- -e - -

Normal antenalal unbllical artary Doppler value Abnormal anten al umbilical artery Doppler value

Fig. 3. Length of stay in neonatal intensive care unit (N/CU) grouped according to antenatal umbilical artery Doppler values in fetuses suspected of having IUGR (n = 90). Asterisk, p < 0.005, very highly statistically significant.

60~----------------------~--------~

50

-----________

20

(28)

---------(21)

(20)

10

O+---------------------------------~ 1-14

Days of Therapy in NICU

Fig. 4. Trends in treatment modalities for SGA infants grouped according to antenatal umbilical artery Doppler values (n = 34). 0, Mechanical ventilation; 6, oxygen therapy; ..... , Doppler value normal; - -, Doppler value abnormal; N/CU, neonatal intensive care unit.

Doppler abnormal groupings. Although birth weight was less in the SGA-Doppler abnormal group com· pared with the SGA-Doppler normal group (1l57 gm vs 1414 gm), these differences were not statistically different. SGA infants with abnormal antenatal Doppler values tended to have a higher frequency of docu· mented hypoglycemia (85% vs 57%, P = 0.07). Al· though intraventricular hemorrhage of any grade is rare in the SGA infants, only one infant (in the abnor· mal group only) had an abnormal head ultrasono· graphic examination. The most distinguishing feature between the two SGA groups classified by abnormal antenatal Doppler studies is the number admitted to the intensive care unit (80% vs 43%, P < 0.05) and duration of special care in the special care unit (38.2 vs

5.5 days, p < 0.05). Fig. 4 shows trends in treatment modalities with both duration of mechanical ventilation and oxygen therapy tending to be longer in the SGADoppler abnormal group, whereas the reverse trend is noted for the SGA-Doppler normal group. Combinations of antenatal umbilical artery Doppler results and neonatal birth weight groups were identi· fied, as noted in Table IV, and analysis of variance statistical tests were applied. Among the four groupings analyzed gestational age was not different and not likely to account for the observed differences. With respect to birth weight, SGA infants with abnormal Doppler reo suIts had lower birth weights compared with SGA babies with normal Doppler results, but these differences were not statistically significant. For birth weight the SGA

Gaziano et al.

Volume 170, Number 6 Am J Obstet Gynecol

Abnormal Doppler A SGA(N_20)

31

Normal Doppler B SGA (N-14)

14

Abnormal Doppler C non-SGA (Nz21)

12

Normal Doppler D non-SGA (N_35)

7

I

1739

P<0.OO5

I I

I

o

40 Mean Length of NICU StaYtf)IIYs)

Fig. 5. Duration of neonatal intensive care unit (NICU) stay (days) on basis of outcome grouping according to antenatal umbilical artery Doppler values (n = 90). Asterisk, Analysis of variance indicates that B, C, and D are not significantly different from each other but that A differs highly significantly from B, C, and D.

Table IV. Relationship among combinations of antenatal umbilical artery Doppler results and neonatal birth weight groupings Variables Groupings

Abnormal Doppler resuits (A), SGA (n = 20) Normal Doppler results (B), SGA (n = 14) Abnormal Doppler resuits (C), non-SGA (n = 21) Normal Doppler results (D), non-SGA (n = 35)

Birth weight (gm)

Mechanical ventilation (days)

Oxygen therapy (days)

1157

24

30

1414

12

20

1591

4

6

1834

6

8

Analysis of variance, Duncan's multiple range test for listed variables. For birth weight A is not significantly different from Band C is not significantly different from D but A is significantly different from both C and D. For mechanical ventilation and for oxygen therapy A is not significantly different from B but A is significantly different from C and D.

babies with abnormal Doppler results differed from those who were not SeA, irrespective of Doppler values. Similar findings are noted for duration of mechanical ventilation and oxygen therapy. Fig. 5 shows that the Doppler abnormal-SeA group had a significantly prolonged length of stay in the neonatal intensive care unit (31 days) compared with the seA normal group (14 days) or to the other two non-SeA groups. Again, the SeA-Doppler normal group did not differ in length of stay compared with non-SeA infants with either normal or abnormal Doppler results. Fig. 6 shows these relationships graphically for duration of neonatal intensive care unit stay. Finally, analysis of variance indicates the length of stay in the special care unit was 38.2 days for abnormal Doppler-SeA infants, statistically different (p = 0.005) from the normal Doppler-SeA infants (5.5 days) and from the non-SeA-abnormal group (13.2 days) and from the non-SeA-normal group (9.7 days). The anal-

ysis showed that the latter three groupings did not differ from each other for length of stay in the special care unit.

Comment Umbilical artery velocimetry appears to be a distinguishing feature for the preterm growth-retarded fetus, identifYing a subgroup destined for lesser or greater risk during the neonatal course. In infants delivered near term (mean 36.6 to 38.9 weeks), Berkowitz et al." noted that seA infants with abnormal Doppler results had increased admission rates and prolonged courses in the neonatal intensive care unit. Reuwer et al. 13 stressed the discriminating value of normal umbilical resistance measures in 19 SeA fetuses, whereas Mulders et al. 7 assessed early neonatal outcome parameters and noted fetal distress, premature delivery, and the presence of seA and low birth weight to be greater in the Doppler abnormal groups. In 15 singleton preterm

1740 Gaziano et al.

June 1994 Am J Obstet Gynecol

50

(SO)

40

Neonates Receiving NICU Therapy ("!o)

30

:

(21)

(21)

20

i (20)

(17)

10 0

>14

1-14 -0- Abnormal umbilical artery DoppierlSGA at birth --6- Normal umbilical artery DoppIerlSGA at bif1!1

-+-

Normal umbilical artery DoppIer/non-SGA at birth

Fig. 6. Length of stay in neonatal intensive care unit (N/CU) grouped according to antenatal umbilical artery Doppler values and neonatal birth weight status (n = 90). 0 - 0 , Abnormal umbilical artery Doppler results, SGA at birth; I'r---f':" normal umbilical artery Doppler results, SGA at birth; - , normal umbilical artery Doppler results, non-SGA at birth.

pregnancies McCowan et al. B reported a significantly increased risk for SGA infants with abnormal Doppler findings, but only one of the patients had a normal Doppler value. Hackett et al. 9 studied flow in the descending fetal aorta and confirmed the increased risk for perinatal death, necrotizing enterocolitis, and hemorrhage, whereas 7S% of SGA infants with normal Doppler values had uncomplicated courses. Groenenberg et al. 14 established that umbilical artery velocimetry is superior to velocimetry of the fetal ascending aorta, pulmonary artery, or internal carotid artery for the prediction of neonatal outcome events. In spite of previous investigations a number of questions remain unanswered. What is the incidence of normal umbilical artery velocimetry in the SGA fetus? What level of abnormality justifies a change in risk for the SGA fetus - zero flow or 2 SDs above the mean? What is the nature of the increased morbidity, and how is it manifested? This study shows the likelihood of a normal Doppler measurement in an SGA neonate to be approximately 39%. This figure compares with S1 % noted by Berkowitz et al. 6 for near-term SGA infants and 44% noted for preterm infants by Hackett et al. 9 These figures reflect the diverse causes for SGA infants but also suggest that management for a significant number of suspected SGA fetuses could be altered if those with normal Doppler results are in fact a separate subgroup. The current study also confirms the value of using ~ 2 SDs above the mean for gestational age as a cutoff value for Doppler abnormality. Only seven patients in this series had no flow in diastole, whereas the remaining 41 patients were ~ 2 SDs above the mean. This

latter standard appears to discriminate degree of risk well, although it is acknowledged that zero-flow neonates are likely to have more serious compromise. 9 The study supports the concept that fetuses at risk for SGA can be classified into two groups on the basis of Doppler abnormality. The risk for prolonged neonatal intensive care unit stay is no greater for the SGAnormal Doppler baby than it is for the non-SGA baby with or without abnormal antenatal Doppler results. Pre term SGA infants with normal Doppler values are admitted to the neonatal intensive care unit half as often as those with abnormal values. A profound difference exists in extended care for SGA infants with normal Doppler results (S.S vs 38.2 days). Overall, infants in this study with an abnormal Doppler result had a higher incidence of intraventricular hemorrhage, but the number of SGA infants with intraventricular hemorrhage is too small to make inferences. SGA infants with normal Doppler values had a higher but not statistically different birth weight compared with SGA babies with abnormal Doppler results. This difference might partially account for the higher frequency of neonatal intensive care unit admissions and prolonged stay for the Doppler abnormal-SGA group. The duration of mechanical ventilation and other therapies was longer in this group but did not always reach significance. Early neonatal outcome parameters such as Apgar score and need for delivery room resuscitation was not different among the different Doppler groups, and this observation has been reported by others.9 These findings are consistent with our policy to avoid acidosis in the fetus suspected of growth retardation.

Gaziano et al.

Volume 170, Number 6 Am J Obstet Gynecol

Abnormal velocimetry in the umbilical artery is thought to reflect increased resistance to flow in the uteroplacental circulation! These changes likely reflect long-term compromise to maternal-fetal gaseous and nutrient exchange and may result in more profound degrees of retarded growth with fewer neonatal reserves in those fetuses with abnormal velocimetry. In this study six of seven neonates with abnormal Doppler results and severe RDS died, indicating that a factor in neonatal survival may be determined by the chronicity of compromise as reflected by abnormal velocimetry. For the surviving SGA fetuses in this study neither gestational age nor birth weight differences accounted for the observed differences/i~ morbidity. This paper would support a 'system for classifying risk in fetal growth retardation in part on the basis of umbilical artery Doppler results. Because growth retardation has multiple causes, morphologic, genetic, and biometric fetal assessment is necessary. Pardi et al. 15 suggested that Doppler velocimetry may distinguish SGA fetuses who are candidates for fetal blood assessment. More important, reassessment of risks for the growth-retarded fetus may allow for fewer interventions and less intensive surveillance for a large number of pregnancies. REFERENCES

1. Fleischer A, Schulman H, Farmakides F, Bracero L, Blattner P, Randolph G. Umbilical artery velocity waveforms and intrauterine growth retardation. AM J OBSTET GYNECOL 1985;92:502. 2. Erskine RLA, Ritchie JWK. Umbilical artery blood flow characteristics in normal and growth retarded fetuses. Br J Obstet Gynaecol 1985;92:605. 3. Gaziano EP, Knox GE, Wager GP, Bendel RP, Boyce DJ, Olson J. Predictability of SGA infants by real-time ultrasound-derived measurements combined with pulsed Doppler umbilical artery velocimetry. AM J OBSTET GyNECOL 1988;158:1431. 4. Trudinger BJ, Cook CM. Umbilical and uterine flow velocity waveforms in pregnancy associated with major fetal abnormality. Br J Obstet Gynaecol 1985;92:666. 5. Doubilet PM, Benson CB. Fetal growth disturbances. Semin Roentgenol 1990;25:309. 6. Berkowitz GS, Mehalek KE, Chitkara U, Rosenberg J, Cogswell BS, Berkowitz RL. Doppler velocimetry in the prediction of adverse outcome in pregnancies at risk for intrauterine growth retardation. Obstet Gynecol 1988;71: 742. 7. Mulders LGM, Jongsma HW, Hein PRo Uterine and umbilical artery blood flow velocity waveforms and their validity in the prediction of fetal compromise. Eur J Obstet Gynecol Reprod Bioi 1989;31:143. 8. McCowan LM, Erskine LA, Ritchie K. Umbilical artery Doppler blood flow studies in the preterm, small for gestational age fetus. AM J OBSTET GYNECOL 1987;156:655. 9. Hackett GA, Campbell S, Gamsu H, Cohen-Overbeek T, Pearce JMF. Doppler studies in the growth retarded fetus and prediction of neonatal necrotising enterocolitis, haemorrhage, and neonatal morbidity. BMJ 1987;294:13. 10. Brar HS, Platt LD. Reverse end-diastolic flow velocity on umbilical artery velocimetry in high-risk pregnancies: an ominous finding with adverse pregnancy outcome. AM J OBSTET GYNECOL 1988;159:559. 11. Burke G, Stuart B, Crowley P, ScanaiII SN, Drumm J. Is

1741

intrauterine growth retardation with normal umbilical artery blood flow a benign condition? BMJ 1990;300:1044. 12. Gaziano EP, Knox E, Bendel RP, Calvin S, Brandt D. Is pulsed Doppler velocimetry useful in the management of multiple gestation pregnancies? AM J OBSTET GYNECOL 1991; 164: 1426. 13. Reuwer PJHM, Reitman GW, Sijmons EA, et al. Intrauterine growth retardation: prediction of perinatal distress by Doppler ultrasound. Lancet 1987;2:415. 14. Groenenberg lA, Bearts W, Hop WC, Wladimiroff JW. Relationship between fetal cardiac and extra-cardiac Doppler flow velocity waveforms and neonatal outcome in intrauterine growth retardation. Early Hum Dev 1991;26: 185. 15. Pardi G, Cetin I, Marconi AM, et ar. Diagnostic value of blood sampling in fetuses with growth retardation. N Engl J Med 1993;328:692.

Discussion

FREDRIK F. BROEKHUIZEN, Milwaukee, Wisconsin. In spite of multiple studies the role of fetal umbilical artery velocimetry is still far from certain. Two prospective, randomized trials did not find significant differences in interventions and perinatal outcome parameters, including length of stay in the neonatal intensive care unit. I. 2 Because many causes of IUGR (genetic, nutritional, infectious) do not necessarily result in fetoplacental circulatory changes, Doppler studies have little or no value in the detection of IUGR as a routine test in the general population.'-5 Fetal biometry, clinical findings, and risk assessment are still imperfect tools for suspected IUGR, as was the case in this study. Abnormal Doppler studies, especially absent end-diastolic umbilical artery flow, have shown to predict fetal compromise and ultimately hypoxia, but not the time frame in which this will occur. Abnormal Doppler studies therefore, by themselves, cannot determine the timing of intervention. Normal Doppler studies in cases of suspected IUGR have been reported to be reassuring,6 especially if congenital anomalies, chromosomal disorders, and infections can be ruled out. This study confirms the reassurance of normal Doppler studies in the preterm fetus suspected of having IUGR. The paper states a figure of 39% of normal Doppler studies in the preterm SGA group. This is based on postnatal birth weight. Prenatally 52% of their study subjects had normal Doppler studies. It is interesting that only 38% of fetuses (growing below the 10th percentile by estimated fetal weight) were below the 10th percentile postnatally. Under these circumstances abnormal Doppler studies predicted 46% of SGA babies. The authors confirmed their hypothesis that normal Doppler studies in cases of suspected preterm IUGR prognosticate the immediate neonatal complications and neonatal intensive care unit stay to be similar to that of preterm neonates of appropriate birth weight. Does this represent "benign" or maybe "constitutional" IUGR? I am not so sure that the other side of the hypothesis (abnormal Doppler studies define a subgroup with increased morbidity) has been confirmed. Abnormal Doppler studies in this paper cannot be correlated with the severity of the IUGR, because differences in neonatal morbidity and neonatal intensive care unit stay DR.

1742

Gaziano et al.

between the SGA-normal Doppler and SGA-abnormal Doppler groups cannot be explained by significant differences in birth weight between these groups. There were no significant differences between the nonSGA-normal Doppler and non-SGA-abnormal Doppler groups in neonatal morbidity and neonatal intensive care unit stay (keep in mind that prenatally all fetuses were suspected to have IUGR), thereby not confirming a predictive value for Doppler studies in suspected IUGR, when the postnatal birth weight is > 10th percentile. I have two questions. Do the data in this paper support a prospective, randomized trial in suspected preterm IUGR with one arm (after normal Doppler studies and exclusion of anomalies) noninterventive with no antenatal testing? Conversely, this paper suggested that the duration of mechanical ventilation and oxygen therapy was prolonged in the SGA-abnormal Doppler group. That could be related to pathologic circulatory adaptations in the fetal pulmonary circulation prenatally. Would the authors support a prospective trial that, in the case of absent end-diastolic flow and IUGR, would randomize between delivery or continued antenatal testing until an abnormal test result by biophysical profile, nons tress test, or oxytocin challenge test was found? REFERENCES 1. Trudinger BJ, Cook CM, Giles WB, et al. Umbilical artery flow velocity waveforms in high risk pregnancy. Randomised controlled trial. Lancet 1987;1:188-90. 2. NewnhamJP, O'Dea MR, Reid KP, Diepeveen DA. Doppler flow velocity waveform analysis in high risk pregnancies: a randomized controlled trial. Br J Obstet GynaecoI1991;98: 956-63. 3. Beattie RB, Dornan JC, Clements CWA, et al. Umbilical artery velocity wave form assessed as an antenatal screening tool for intrauterine growth retardation and poor fetal outcome. Br J Obstet Gynaecol 1988;95:534-9. 4. Berkowitz GS, Mehalek KE, Chitkara U, et al. Doppler umbilical velocimetry in the prediction of adverse outcome in pregnancies at risk for intrauterine growth retardation. Obstet Gynecol 1988;71:742-6. 5. Dempster J, Mires GJ, Patel N, et al. Umbilical artery velocity wave forms: poor association with small for gestational age babies. Br J Obstet Gynaecol 1989;96:692-7. 6. Reuwer PM, Rietman GW, Symons EA, et al. Intrauterine growth retardation: prediction of perinatal distress by Doppler ultrasound. Lancet 1987;2:415-8. DR. DANIEL A. RIGHTMIRE, Springfield, Illinois. Numeric data gleaned from the tables and figures were analyzed to assess the efficacy of an abnormal umbilical artery SID ratio as a test for the abnormal neonatal outcomes selected by the authors. In this report the Doppler test performed very poorly. The data in Table I are interpreted as evidence that the SID ratio was a poor test for SGA and for neonatal death. The K values, which are estimates of the overall agreement between the test results and the actual presence or absence of disease corrected for the expected agreement by chance, indicated low levels of association. Similarly, the SID test performed very poorly as a predictor for

June 1994 Am J Obstet Gynecol

neonatal intraventricular hemorrhage and requirement for intravenous therapy. In fetuses actually born SGA the Doppler studies as a test for hypoglycemia also performed very poorly. The efficacy of the SID ratio as a test for newborn intensive care unit admission had a demonstrably poor performance. This occurred in all three groups tested, whether the SGA condition was suspected or actually present or actually not present. The SID ratio was .also a poor test for prolonged neonatal intensive care unit admission in all three groups of patients in spite of the fact that 27%, 38%, and 20% of the fetuses, respectively, stayed in the neonatal intensive care unit for > 14 days. The study confirmed an association between neonatal morbidity and abnormal umbilical artery Doppler findings. Nevertheless, the associations would appear to be of limited clinical utility, because poor results are predicted if clinicians presumed to act on the results. The authors argue that the SID of > 2 SDs above the mean for gestational age is a better cutoff point than is the absence of end-diastolic velocities. The appropriate argument must be based on receiver-operator characteristic curves, which have not been presented. 1 In general, of all possible cutoff values the best performance is characterized by the greatest number of truepositive results and the least number of false-positive results, although sometimes clinicians will choose the value that detects all extremely bad outcomes. If the absence of end-diastolic velocities is associated with excessive morbidity, the abnormal SID ratio only slightly > 2 SDs over the mean does not appear to be associated with enough morbidity for it to be useful clinically. As a predictor of umbilical venous hypoxemia the SID ratio performed poorly until the end-diastolic velocity was reduced to the point where the SID ratio exceeded 4.5. 2 In this study the likelihood ratios were in the 1.6 to 2.7 range for the outcome variables (neonatal conditions) in Table I. Thus the data do not support the argument that the SID ratio is a good test for neonatal morbidity. Nor has the argument been made convincingly that the appropriate cutoff value for best performance had been chosen. Nevertheless, data in our literature suggest that the authors are indeed correct with the thesis suggested by the title of this report. The logical follow-up study to this work would be a randomized, clinical trial of fetuses suspected by ultrasonography to have restricted growth in which little or no intervention would be permitted in the group with normal Doppler ultrasonography studies of the umbilical artery. Such a trial produced data from which we can conclude that the use of Doppler surveillance before labor resulted in statistically and clinically significant reductions in the use of antepartum fetal monitoring techniques, antenatal hospital admissions, labor inductions, and emergency cesarean deliveries for fetal distress, these without differences in gestational age at birth, birth weight, low Apgar scores, or the need for respiratory support. 3 The use of um-

Gaziano et al.

Volume 170, Number 6 Am J Obstet Gynecol

1743

Table I. Clinical efficacy of SID ratio > 2 SDs as a test for neonatal morbidity and mortality Neonatal factor

False positive (%)

Negative predictive value (%)

K

SGA « 10th percentile) Neonatal death Intraventricular hemorrhage Intravenous therapy Hypoglycemia in SGA fetuses Neonatal intensive care unit admission SGA suspected SGA present Not SGA Neonatal intensive care unit stay > 2 wk SGA suspected SGA present Not SGA

40 46 42 25 33

71 94 94 37 43

0.38 0.19 0.13 0.20 0.29

25 33 21

61 57 66

0.42 0.40 0.35

39 48 36

82 79 83

0.20 0.26 0.07

bilical artery Doppler velocimetry in SGA fetuses allowed antenatal monitoring and obstetric interventions to be aimed more precisely then did nonstress testing. REFERENCES 1. Maulik D, Yarlagadda P, Youngblood]P, et al. Comparative efficacy of umbilical arterial Doppler indices for predicting adverse perinatal outcome. AM] OBSTET GYNECOL 1991;164: 1434-40. 2. Tyrrell S, Obaid AH, Lilford RJ. Umbilical artery Doppler velocimetry as a predictor of fetal hypoxia and acidosis at birth. Obstet Gynecol 1989; 74:322-36. 3. Almstrom H, Axelsson 0, Chattingius S, et al. Comparison of umbilical artery velocimetry and cardiotocography for surveillance of small for gestational age fetuses. Lancet 1992;340:936-40. DR. JOSEPH M. MILLER, JR., New Orleans, Louisiana. I would like to know the difference between an intrauterine growth-retarded baby and a baby that's small for gestational age and how those definitions apply to the study. DR. GAlIANO (Closing). First, with respect to Broekhuizen's incisive and cogent comments regarding the role of fetal Doppler, I fully agree. To answer his first question, the next study needs to be prospective, eliminating fetuses suspected of chromosomal abnormalities and congenital defects; those remaining could then be randomized according to Doppler normality or abnormality. We would then be able to answer some of the questions that were raised today. Second, with respect to the issue of zero diastolic flow and to focus on some of Rightmire's comments, we had only seven

patients with zero diastolic flow in our series. If you await that abnormality, it would be very difficult to set up a randomized, double-blind study with such numbers. Currently, patients with zero diastolic flow are followed up expectantly, and we manage them with standard biometry and biophysical profile fetal testing. With respect to Rightmire's comments and statistical evaluation, it was not the purpose of the study to predict growth retardation. We have performed other studies related to growth retardation prediction. (The study numbers were too small to evaluate neonatal death and the other noncorrelations are discussed in the article.) Our definition of entry into the study was based partly on the ultrasonographically estimated fetal weight, but it was also related to physician suspicion of growth retardation. Antenatally, we did not have absolutely tight criteria for patient selection related to defining growth retardation. As Rightmire suggests, with the current numbers I would not have expected SGA prediction to be good. In terms of the differences in terminology for IUGR and SGA, this area certainly is confusing. We prefer to define growth retardation prenatally by the same measures we do neonatally, that is, by percentile birth weight. If the estimated fetal weight is :5 10th percentile, we define that as IUGR prenatally. We tend to refer to those born with a birth weight of :5 10th percentile postnatally as SGA. (We acknowledge that not all SGA infants are growth retarded.)