Antenatal corticosteroids to prevent neonatal respiratory distress syndrome: Risk versus benefit considerations

Antenatal corticosteroids to prevent neonatal respiratory distress syndrome: Risk versus benefit considerations RICHARD JOHN

J. BOEHM,

JANET SHARON JOHN

DEPP, A. NOSEK,

M.D. M.D. B.S.N.

L. DOOLEY, M. HOBART,

M.D. M.D.

Chicago, Illinois The clinician considering administration of steroids to prevent respiratory distress syndrome (RDS) should attempt to identify patients who do not have criteria previously shown to increase the likelihood of benefit in the prevention of RDS. it is possibJe to accurately predict the intarvai to delivery in most cases. Four hundred thirty-nine patients at risk to deliver prior to 37 weeks have been screened for factors known to decrease the iiketihood of benefit. Only 47 (10.7%) screened candidates have no exclusion criteria. Twenty-seven (6.9%) of 392 excluded neonates deveioped RDS; 20 of the 27 were predicted to and did deliver in less than 24 hours after initial screening. Oniy one case was inappropriately excluded. Fetal surfactant assessment is crucial; puimonaly maturity excludes 19% at 28 to 33 weeks and 36% at 34 to 37 weeks. Careful screening for exclusion factors known to decrease the likelihood of steroid derived benefit is essential when use of a drug with potential long-term consequences (risk) is considered. (AM. J. OBSTET. GYNECOL. 137238, 1980.)

DISTRESS SYNDROME (RDS) continues to be the leading cause of neonatal death. Farrell and Wood’ report that approximately 12,000 neonates per year die of hyaline membrane disease (HMD/RDS) thus establishing the magnitude of the problem. In 1969, it was estimated that approximately 25% of neonates born prematurely show signs of RDS and 20% to 40% eventually die.2 In 1973, Klauss3 estimated that from 0.5% to 1.0% of all infants delivered die from this disease. Prevention of physician-induced RDS by using amniotic fluid surfactant assessment and early pediatric support utilizing continuous distending airway pres-

RESPIRATORY

From the Division of Obst&ics, Northwestern University Medical School, and Prentice Women’s Hospital and Maternity Center of Northwestern Memorial Hospital. Presented at the Forty-seventh Annuul Meeting of the Central Association of Obstetricians and Gynecologists, White Sulphur Springs, West Virginia, September 21-23, 1979. Reprint requests: Dr. Richard De@, Director, Division Obstetrics, Prentice Women’s Hospital and Maternity Center of Northwestern Memorial Hospital, 333 East Sufmior St., Chicago, Illinois, 6061 I.

338

of

sure has undoubtedly made an impact on mortality data. Perhaps the most exciting mode of therapy has been generated by studies in both animals and humans suggesting that the administration of corticosteroids (CS) to the mother 24 hours prior to delivery may reduce the severity and incidence of RDS. All studies to date support the conclusion that steroids are beneficial. Animal studies lay the foundation to support the suggestion that prenatal CS hasten fetal lung development. 4-8 Liggins and Howieg provided the first human data confirming the relationship in 1972. Liggins enlarged his series in 1977 with similar conclusions.‘O Since 1972, several authors*1-‘5 have reported, on the basis of smaller and less well-controlled human series, that CS have benefit in reducing the incidence of RDS. More recently, several larger series have also supported this relationship.1”20 The study by Block and associatesI is of interest in that it provides information regarding the impact of pretreatment assessment of the lecithin/sphingomyelin (L/S) ratio. It is important to note that, to date, the study of Liggins has not been verified by a similar large study which consid0002-9378/80/l

10338+13$01.30/O

0 1980 The C. V. Mosby

Co.

Antenatal cotticosteroids

ers other variables known to modify the time of appearance in gestation of surfactant or the impact of surfactant assessment on patient selection.21 Until such a study is accomplished, the clinician is limited to conclusions which can be derived from the observations made to date. These observations are largely confined to characteristics of patients who apparently derived benefit from steroid therapy in Liggins’ study. Strong consideration of the risk to benefit ratio is essential prior to the prenatal administration of any medication. The potential for either short-term or long-term consequences of antenatal steroids is great; this subject has been reviewed extensively elsewhere.22-25 The lung, brain, liver, gut, kidneys, enzyme systems, immune system, and pancreas are all potential targets of side effects. In addition, Tauesch and associates’$ have recently indicated an increased risk of maternal infection after steroid administration. This has not been supported by other studies which report no additional hazard.12s 2o Realistic assessment of the risk to benefit ratio is not possible without objective, realistic assessment of the risk of RDS and the likely impact of agents such as steroids on the individual patient. In the most recent report by Liggins, two factors were of prime importance in assessing potential benefit.‘O Gestational age less than 34 weeks, particularly 30 to 32 weeks, and steroid administration to delivery interval (SADI) of 25 hours to 7 days are factors predicting maximum benefit. Ballard and associates18 suggest that the interval must be greater than 48 hours. Antenatal assessment of the L/S ratio is clearly an important third variable. Block and associates16 report a significant increase in benefit between control and CS treated groups when the L/S is known to be less than 2.0. Thus, it appears that the ability to correctly predict that delivery will be delayed longer than 24 hours but less than 7 days, to predict gestational age accurately, and where possible, to perform amniocentesis to reduce the likelihood of CS administration to a premature fetus with mature lung function are key variables in potential CS benefit assessment. Gluck and associates20s 21 have indicated that one may attain an L/S ratio greater than or equal to 2.0 in a significant proportion of fetuses less than 32 weeks or in the 800 to 1,200 gm weight range. Previous literature seems to imply that pregnancies at risk to deliver prior to 37 weeks can be divided into three groups. Significant difference in outcome occurs only in patients with maximum benefit characteristics (MaBC). Patients in the maximum benefit group (MaBG) have MaBC which include fetuses less than 34 weeks’ gestation and a SADI greater than 24 hours but less than 7 days.‘O

to prevent RDS

339

Very little data are available regarding a second group of pregnancies with minimum benefit characteristics (MiBC). This group includes those patients of greater than 34 weeks’ gestation with unknown surfactant activity. Clearly, there are individual cases where benefit may occur; however, patient selection is a key factor because of the higher frequency of mature lung function at 34 to 37 weeks’ gestation. The incidence of L/S ratio less than 2.0 in such a population provides an estimate of the percentage of patients who might derive benefit; however, not all neonates with a prior L/S less than 2.0 will develop RDS; the impact of an immature L/S ratio may differ when one compares neonates at 28 to 33 weeks’ versus 34 to 37 weeks’ gestation. The third group of patients are those with no benefit characteristics (NBC). This group includes those pregnancies which are ineligible for steroid therapy (see Material and methods section) and those with known (mature L/S), or present but unsuspected, fetal pulmonary maturity (no amniocentesis performed). Also included in this group are pregnancies that will deliver in less than 24 hours or greater than 7 days. Patients delivering in those intervals after receiving st.eroids are patients in which the risk (potential side effects) exceeds the benefit. It is important that data regarding the likelihood of improperly predicting a SAD1 of greater than 24 hours be developed to establish the magnitude of the population to be treated which will include most patients who will actually deliver wit.hin the ideal SADI. In the series of Liggins, 159 (18.6%) and 356 (41.7%) of 853 patients in the total series were delivered at less than 24 hours or greater than 7 days from the time of steroid administration. Both are outside the ideal SADI.” It is not possible to determine the ability to correctly predict this interval from Liggins’ data since prediction of such an interval was not an objective of his study. Nonetheless, this (60.3%) represents an estimation of the population of premature infants unlikely to derive benefit from antenatal CS administration simply because of physician inability to predict the interval. All neonates less than 37 weeks do not develop RDS. If one evaluates Liggins’ entire control group, 37 of 156 (23.7%) develop RDS. However, steroids evidently do not have a beneficial effect on the patients at 34 to 37 weeks delivering in the ideal SAD1 in the absence of data regarding fetal surfactant activity. In the control group less than 34 weeks with MaBC, 33 (40.2%) of 82 (MaBG) developed RDS. Therefore. theoretically, at least 59.8% derive no benefit. The reduction of RDS incidence between control and steroid groups with MaBC is 40.2% to ll.O%, a reduction of.‘72.6%. The

340

Depp

et al.

Table I. Incidence of exclusion candidates screened for steroid

criteria therapy

Anticipated < 24 hours Anticipated > 7 days L/S ratio 2 2 Cervix 2 5 cm Patient refusal Suspect infection Prior steroids Physician refusal Other

43.3 35.6 7.2 7.2 6.7 2.5 1.5 0.5 9.3 iiiiG---

‘Total patients

84 69 14 14 13 5 3 1 18 194

Note: Some patients have more

than

one

among

122 50 27 20 3 3 4 0 8 198 exclusion

439

Material and methods

61.6 25.3 13.6 10.1 1.5 1.5 2.0 0 4.0 100.0 factor.

population of neonates less than 34 weeks’ gestation who derive benefit from CS can be better estimated by subtracting the incidence of RDS in CS group from the control group. Thus, if the control group incidence is 40.2% and the CS group is 1 l.O%, 29.2% derive benefit; the remaining 70.8% derive little or no benefit. The risk to benefit ratio is thus approximately 2.4 (70.8 to 29.2) for all neonates less than 34 weeks if no data regarding fetal surfactant activity are available. Similar calculations can be established for intervals of less than 30 weeks and inclusive intervals of 30 to 31 and 32 to 33 weeks; in those intervals, 29.9% 47.3%, and 12.9%, respectively, derive benefit. The risk to benefit ratio for each interval is thus 2.3, 1.1, and 6.8, respectively. Clearly the benefit is maximal in the 30 to 31 weeks inclusive interval. Thus, if one were able to predict the ideal SAD1 with 100% accuracy, determine fetal surfactant activity in all cases, and develop a technique of surfactant assessment that predicts RDS with 100% accuracy, as well as establish gestational age accurately in all cases, approximately 50% of the 30 to 31 weeks neonates would derive benefit from antenatal CS administration. Obviously, this is not possible; thus, it is important to determine the accuracy of clinical judgment in these areas so as to determine the magnitude of the population less than 34 weeks that must receive CS to reach the population percentage that will actually derive benefit. To the contrary, it is important to know how many cases will be missed, and thus not derive benefit, because of physician error in prediction of the ideal SADI. Prospective

evaluation

of a population

of patients

at

risk to deliver prior to 37 weeks to assess the impact of gestational age, ability to predict the admission and/or steroid administration to delivery interval, and L/S ratio has been performed to provide ter the assessment of a realistic risk antenatal

CS therapy.

data that to benefit

will ratio

betfor

Patients included in this study were at risk to deliver prior to 37 weeks’ gestation between June 1. 1977 anti February 15, 1979. Four hundred thirty-nine patients between 28 and 37 weeks’ gestation were screened foipresence of criteria which would contraindicate steroid therapy. (All patients with no contraindications were candidates for steroid therapy: results of their man agement and outcome will be presented at a later date.) TO be eligible for consideration for steroid therapy, patients had to be a high risk for delivery between 28 weeks and 36 weeks and 6 days’ gestation from the first day of their last menstrual period. All patients at risk for delivery in that interval were screened upon admission. In the case of inaccurate menstrual period, the assignment of gestational age was determined by the interaction of clinical and ultrasound data. Patients were considered ineligible for steroid therapy if (1) the estimated time to delivery was less than 24 hours; (2) the estimated time to delivery was greater than 7 days; (3) there was evidence of intrauterine infection; (4) the cervix was ~5 cm dilated; (5) the patient was receiving or had received CS at any time during pregnancy; (6) the patient had a history of peptic ulcer disease, active tuberculosis, or viral keratitis; (7) there were known severe fetal anomalies; (8) consent could not be obtained or the patient was unable to comprehend consent; (9) the patient was included in our Diabetes in Pregnancy Center study; or (10) it was likely that the neonate would be unavailable for followup. Determining gestational age at the time of screening. If the patient was a reliable historian and (1) had a history of regular menstrual periods; (2) had not taken oral contraceptives for 3 months prior to her last menstrual period; (3) the date of her last menstrual period could be accurately determined; and (4) the usual pregnancy landmarks could be identified (onset of fetal movement, onset of fetal heart tones, correlation with fundal height measurements), the patient was then said to have reliable dates. The gestational age was then determined on the basis of the time elapsed since the first day of the last menstrual period. If such criteria foiassignment of gestational age were not available, the patient was said to have unreliable dates; the gestational age was then determined with the assistance of sonographic estimation of the fetal biparietal diameter. Since a large proportion of patients at Prentice Women’s Hospital have routine ultrasound at 20 to 26 weeks’ gestation, ultrasound was relied on to a considerable extent in confirming gestational age. Ultrasound determinations after 28 weeks’ gestation were considered less reliable.

Volume Number

I Y7 3

Antenatal

Table II. Accuracy of prediction of steroid administration at less than 34 and at 34 to 37 weeks’ gestation No. of patients

to delivery

in each category

interval

by actual

interval

gestatzon

4-7

8-14

34-37 (3-k

103 73 176

10 6 16

4 1 5

2 3 5

0 0 0

3 1 4

34-37 <34

4 8 12

6 15 21

2 7 9

5 6 11

7

2 .5 7

34-37 <34

3 1

2 3

1 4

z

10 11

4

5

5

4

Total

34-37

110

18

7

9

Total

<.3-k

82

24

12

192

42

19

Total ru Total

Total Note:

Actual

interval

expressed

in hours

up to 72 hours;

predicted

341

Days 49-72

2-7

RDS

to delimy

25-48

Total

to prevent

patients

o-24

O-24

Wrrki’

in excluded

internal

HOW5 Predicted

corticosteroids

interval

215

Tota 1

1%

122 x4 206

61.6 x9.4 64.0

26 4 I C-7

13.1 “1.1 20.x

32 48

50 69

25.3 35.6

21

80

119

37.0

17

37

19x

100.0

11

11

54

I94

100.0

20

28

91

392

100.0

is in hours

!:

up to 24 hours.

All other

intervals

are in

days.

A shake test was performed when amniotic fluid specimens were obtained. If the results were strongly positive, this was considered to be an indication of pulmonary maturity; such a patient regardless of gestational age is ineligible for steroid therapy. An L/S ratio was performed on all patients whene,ver possible; a second L/S ratio was encouraged in cases undelivered on the seventh day after steroid therapy. Patients at less than 34 weeks’ gestation were not required to have an L/S ratio unless there was suspicion that fetal pulmonary maturity might be advanced as a function of maternal complications such as hypertension, proteinuria, placental insufficiency, drug addiction, or methadone or other narcotic use. Such patients required amniocentesis for L/S ratio assessment prior to steroid therapy. However, where feasible, an attempt was made to obtain amniotic Huid in all patients less than 34 weeks to rule out the possibility of present but unsuspected pulmonary maturity. More stringent criteria were applied to patients at 34 to 37 weeks’ gestation. In this group, steroids were considered only if an L/S ratio could be obtained which indicated lack of fetal pulmonary maturity. Prediction of the interval to delivery in all cases was subjective; a score to predict the interval is not available. Prediction of an interval less than 24 hours was relatively easy in most cases. Many patients were admitted with advanced cervical dilatation (4 to 5 cm); many were maternal transports from outlying Level II Centers who received uterine tocolysis in route to our center. Others failed to respond to high dose isoxsuprine and made rapid progress shortly after admission. Simi-

larly, a number presented with premarule rupture of the membranes with some bleeding; most such patierits delivered within 24 hours. Patients who had uterine contractions but Irssc~~ degrees of cervical dilatation (less than ~~cm). thosr who did not demonstrate stretch on the cervix with contumacy tions, those whose lower uterine segments werr not ballooned out, or those who responded easih and immediately to tocolysis attempts were predictcAd to delivrl after 24 hours. Many patients predicted IO deliver at greater than 7 days were medically complicated patients who were admitted for acute ca~.e and at risk for premature intervention in whom marernal status stabilized under observation and NSTIOCX testing \+as reassuring. Others were patients with premature rupture of the membranes at 24 to 30 weeks with Ci long, closed cervix in whom a passive “wait and scv” appreach is our current management scherr~c. All patients with premature labor or wirh pvctnaftwt’ rupture of the membranes with coincicletrral labor IF ceived isoxsuprine in an attempt to stop labor fcor- at least 48 hours, unless there was a contr;~~ndicatic,n to tocolysis, if fetal pulmonary immatm ity ‘it as suspected. Isoxsuprine was not given to patient\ su
342

Depp

et al.

Table III. Accuracy of prediction of steroid administration to delivery interval in patients predicted to delivery in 2 to 7 days (valid candidates) at less than 34 and 34 to 37 weeks No. ofpatients

34-31 34 Total %

0 6 6 12.8

2 11 13 27.7

2 3 5 10.6

in

each category

; 13 27.7

2 1 3 6.4

0 7 7 14.9

;z 47

25.5 74.5 100.0

Note: Actual interval expressed in hours up to 72 hours; predicted interval is in hours up to 24 hours. AI1 other intervals are in days.

newborns ineligible for potential steroid therapy benefits. This group includes patients with both MiBC and NBC. The data derived from the study of these patients are of value in better assigning risk to benefit considerations of antenatal CS therapy. Four hundred thirty-nine patients between 28 and 36 inclusive weeks were screened as potential candidates for antenatal steroids. Forty-seven (10.7%) met the criteria for steroid therapy. The remaining 392 patients (89.3%) were considered ineligible for steroid therapy but were nonetheless at risk for RDS. One hundred ninety-eight (50.5%) of the excluded patients were at least 34, but less than 37, weeks’ gestation; 194 (49.5%) were less than 34 weeks’ gestation.

Results Outcome according to predicted screening to delivery interval. Table I summarizes the reason(s) for exclusion of 392 patients excluded from steroid therapy. Most were excluded because their anticipated screen to delivery interval was less than 24 hours or more than 7 days. Seventy-nine percent of all excluded patients at 28 to 33 weeks and 87% at 34 to 37 weeks had one of these MiBC. Forty-seven patients were considered to be candidates for steroid therapy and thus predicted to deliver in the ideal SADI; that is, they had a predicted screen to delivery interval of 24 hours to 7 days. Thirty-five patients (74.5%) were at less than 34 weeks, and 12 (25.5%) were at 34 to 37 weeks. Thus, if one evaluates the total 439 patients screened as potential candidates for steroids, 229 patients less than 34 weeks were screened; 194 (84.7%) were excluded; 79% of the 194 excluded patients were excluded because the screen to delivery interval was not consistent with the ideal SADI. Two hundred ten patients were at risk to deliver at 34 to 37

weeks; 198 (94.3%) were excluded. Eighty-seven pclcent of the 198 excluded patients were anticipated to deliver outside the ideal SADI. Two factors are most responsible for the difference in exclusion rates for the less than 34 weeks versus the greater than 34 weeks. More patients in the 34- to 37. week category were predicted not to deliver in the ideal SADI. Of patients in the 34 to 37 week category 7.2% were excluded from consideration of steroid therapy because of a mature L/S ratio; 13.6% were excluded in the less than 34 week category. Accuracy of screen: delivery interval prediction. Prediction of the ideal SADI is obviously a crucial clinical issue. Failure to correctly predict that a neonate will deliver in the ideal SAD1 may deny it the advantage of steroids; in contrast, failure to correctly predict the less than 24 hour interval increases risk of side effects without benefit of steroid. Patients can be categorized in one of four groups according to whether the prediction to deliver in 1 to 7 days is correct or incorrect (truepositive versus false-positive) and whether the prediction that the neonate will deliver outside the 1 to 7 day interval is correct or incorrect (true-negative versus false-negative). The false-negative category includes two groups, those delivering in the 24 hour to 7 day interval and those greater than 7 days. The falsenegative group, particularly those delivering in the 24 hour to 7 day interval, is of major concern; this group of neonates may be denied the potential advantage of steroids because of physician error. False-positive prediction of an ideal SAD1 includes two groups, those delivering in less than 24 hours and those delivering after 7 days. False-positive prediction, particularly those delivering in less than 24 hours, increases the risk to benefit ratio. Tables II and III summarize our experience in predicting the correct SAD1 interval, in both the steroid and excluded groups. Data is available only for patients in whom the predicted interval was a factor in the decision process. One hundred seventy-six (85.4%) of 206 excluded patients (Table II) predicted to deliver in less than 24 hours did so (true-negative). Twenty-six (12.6%) of 206 were incorrectly predicted to deliver in less than 24 hours but delivered in the desired interval (false-negative), thus reflecting the relative magnitude of the population who might derive benefit from steroids, but are incorrectly excluded. An additional four patients did not deliver for greater than 15 days even though the predicted SADI was less than 24 hours. Steroid therapy and increased gestational age may both contribute to lung maturation in this small group. All patients predicted to deliver in 2 to 7 days with no other exclusion criteria are candidates for steroid ther-

Volume

137

Number

3

Antenatal cotticosteroids

Table IV. Actual screen to delivery 34 and 34 to 37 weeks’ gestation

interval

in excluded

patients

I

by problem Actual

category

at less than

Days

25-48

49-72

4-7

8-16

>I4

Total

38 44 82

9 7 16

3 2 5

4 4 8

2 3 5

17 19 36

73 79 152

22 33 55

10 5 15

4 4 8

5 0 5

2 4 6

10 4 14

53 50 103

Both <34 234 Total

9 21 30

3 1 4

3 1 4

0 1 1

0 1 1

2 0 2

17 25 42

Hypertension <34 234 Total

8 8 16

3 5 8

0 0 0

0 3 3

3 4 7

5 8 13

19 28 47

Other <34 z34 Total

4 5 9

1 0 1

2 0 2

2 1 3

4 5 9

20 6 26

33 17 50

Total* %

192 49

44 11.2

19 4.8

20 5.1

28 7.1

91 23.2

Premature <34 234 Total

labor

Premature <34 234 Total

rupture

*Total

of 394

reflects

C.24

343

intewal

HOWS Problem and gestational interval

to prevent RDS

of membranes

multiple

problems

for

each

patient.

Percent

apy. Some patients, however, will have exclusion criteria which are independent of the predicted interval. Nonetheless, evaluation of the total group provides useful information regarding the ability to predict the correct 2 to 7 day interval. Forty-one (61.2%) of 67 excluded patients (Table II) were correctly predicted to deliver in the 2 to 7 day interval (true positive), and 26 (38.8%) of 67 delivered outside (false positive) the predicted ideal SADI. However, only 12 (17.9%) delivered in less than 24 hours; this small group has a higher risk to benefit ratio. All patients in the steroid group were predicted to deliver in 24 hours to 7 day,s (Table III). Six (12.8%) delivered in less than 24 hours (false positive); 10 (21.3%) delivered at greater than 7 days. Thirty-one (66%) were true positive cases. Thus, in the total population of 439 patients, 206 (46.9%) were excluded because the predicted interval was less than 24 hours; 176 were true negatives. One hundred fourteen (26%) were predicted to deliver in 1 to 7 days. Seventy-two (63%) were correctly predicted (true positive); unfortunately 41 of the 72 were excluded for other reasons. Forty-two (36.8%) were incorrectly predicted to deliver in the ideal SAD1 (false positive). Only 18 (15.8%)

calculated

on base

of 392

actual

excluded

394 patients.

delivered at less than 24 hours and thus, if given steroids, would have received no benefit. Actual delivery interval. Table IV summarizes the data regarding actual SAD1 in both excluded and study groups. Such data, reflects the best that one can attain with perfect prediction of delivery interval. It is thus an estimate of the percentage of patients by category who can be expected to derive benefit from steroid administration and patients (either less than 24 hours or greater than 7 days) in whom risk versus benefit ratio may be less favorable. In the excluded group, 49% of patients delivered in less than 24 hours; 21.1% delivered in the ideal SAD1 (Table IV). Delivery was delayed greater than 7 days in 30.3% of patients. In the steroid group, 12.8% delivered in less than twenty-four hours; this reflects patient selection (Table V). Sixty-six percent (66%) delivered in the ideal SADI, while 21.3% delivered after 7 days. It is interesting to note that patients with premature labor or premature rupture of the membranes (PROM) accounted for all false positive cases; labor inhibition was unsuccessful in all six. Assessment of pulmonary maturity. An L/S ratio was obtained (Table VI) in 137 of 439 (31.2’j10) candidates for the study. Sixty-three (28%) of 226 patients at

344

Depp

et al.

Table V. Actual drug/placebo to delivery interval in valid steroid category at less than 34 and 34 to 37 weeks’ gestation

I

candidates

Actual

at less than 37 weeks by problem

intervul

Ho-U-S Problem and gestational interval

Premature

<24

25-48

49-72

4-7

8-14

>I4

Total

2 0

2

0

0

0 1

5 0

9 4

3 0

6

2 1

6

1

0 1

2 0

19

1 0

0 0

0 0

0 2

0 0

0 0

0 0

2 0

1 0

0 0

0 0

4

0 0

1 0 13

0 0

1 0

0 0

2

3 6.4

7 14.9

labor

<34 234

Premature

Days

rupture

of membranes

<34 234

4

Both <34 234

Hypertension

1 2

and PIH

<34 234

1

Other <34 234

Total

6 12.8

5% PIN: Pregnancy-induced

27.7

5 10.6

0

1 13 27.7

1 47

hypertension.

28 to 33 weeks and 74 (35%) of 213 at 34 to 37 weeks had fetal surfactant assessment data available to assist assignment to excluded or steroid groups. Twelve of 63 patients (19%) at 28 to 33 weeks, and 26 of 74 patients (35%) at 34 to 37 weeks had an L/S ratio greater than or equal to 2.0 and thus were not candidates for steroids. Impact of actual screen to delivery interval and L/S ratio in neonatal RDS. The data of the steroid treated patients will be reported in another paper. We will report data primarily related to excluded patients, because the data of excluded patients is important in evaluating the impact of the three variables: gestational age, actual SADI, and L/S ratio in neonatal RDS. It also provides data regarding the impact of not offering steroids. The data for 43 pregnancies with known amniotic fluid L/S ratio at risk to deliver between 28 to 33 weeks is presented in Table VII. The L/S ratio was greater than or equal to 2.0 in 12 of 43 patients (48.8%) at 28 to 33 weeks; they were not candidates for steroid therapy. One of the 12 developed moderate RDS; the amniotic fluid was meconium stained, but the L/S was reported as 2: 1. Four of 31 patients (12.9%) with an L/S ratio less than 2.0 developed RDS; all were delivered within 7 days of the amniocentesis. Four of 20 patients (20%) with an L/S ratio less than 2.0 delivering within 7 days at 28 to 33 weeks developed RDS. Delivery was delayed greater than 7 days in 15 patients

(34.9%) in this group, accounting for the overall low incidence of RDS. The data for pregnancies at risk to deliver at 34 but less than 37 weeks is presented in Table VIII. Twentyfour of 61 patients (39.3%) at 34 to 37 weeks had an L/S ratio greater than or equal to 2.0 and thus were not candidates for steroid therapy. One of these 37 patients (2.7%) with an L/S less than 2.0 developed RDS. In the entire group at risk to deliver in the interva1 at 34 to 37 weeks only one (1.6%) developed RDS. Delivery was delayed greater than 7 days in 21 patients (34.4%) which undoubtedly had a beneficial effect. Exclusion criteria for neonates with RDS. To determine whether physician error was responsible for failure to offer steroids, we examined the exclusion criteria for neonates who developed RDS (Table IX). Twenty patients at 28 to 33 weeks and 7 at 34 to 37 weeks were excluded from therapy because the predicted SADI was less than 24 or greater than 7 days. Seven patients could not be treated because of inability to obtain patient consent; one was receiving ethanol inhibition early in the study before isoxsuprine inhibition became standard procedure. Accuracy of interval prediction in neonates with RDS. Inaccurate assignment to the less than 24 hour interval (false-negative) potentially may deny steroid benefit. Review of 27 neonates (Table X) originally screened at 28 to 33 weeks’ and 34 to 37 weeks’ gesta-

Antenatal corticosteroids to prevent RDS

Table VIII. Outcome versus L/S by amniocentesis: Delivery interval in excluded patients screened at 34 to 37 weeks. Interval expressed as hours up to 24 and as days thereafter

Table VI. L/S ratio data in 137 of 439 candidates screened between 28 and 37 weeks LIS

ratio

28-33

I

Cl.0 1.011.4 1.5/1.9 22.0 Total done Not done Total patients

3 27 21 12 63 163 226

weeks (5%) (43%) (33%) (19%) (28%) (72%)

Note: Percent calculations population;

137 patients

I

for

had

34-37 0 20 28 26 74 139 213

weeks (0%) (27%) (38%) (35%) (35%) (65%)

L/S ratios exclude L/S ratios determined.

“not

Patients Interval

done”

(No.

with RDS)

by ir&rual

Intmul

<24

l-7

L/S ratio 0.0-1.4 1.5-1.9 22.0

2 5 cl)* 2(l)t

g(2)* 4(I)* 6

3 1 3

4$ 3s 111

13 (2)

Total

9

19 (3)

7

8

43 (5)

*Severe

8-14

>14

L/S ratio 0.8- 1.4 1.5-1.9 22.0

Total

Table VII. Outcome versus L/S by amniocentesis: Delivery interval in excluded patients screened at 28-33 weeks. Interval is expressed in hours up to 24 and as days thereafter Patients

23 days

to 38, 37, and

b anterual

<24

2-7

8-14

1 3 9

4 14(l)* 9

2 5 2

27

9

I

>14 -I-----

T&d

?I-:3:j 49

12 (0) 25 (1) 24 (0) 61 (1)

12

RDS (34 weeks at delivery). of 30, 26, and 31 days to 38, 37, and

29 weeks’

gestation at delivery (no delivery data in two. maternal transports). $Intervals of 26,47, and 20 days to 39.42. and 38 weeks at delivery. glntervals of 28, 26, 23, and 34 days to 38. 38, 39 and 41 weeks at delivery.

IX. Exclusion

criteria

18 (2)

37 weeks.

/IMaternal transport. Delivery data not available, referred back to hospital of origin. tion respectively, reveals that 14 (70%) and 5 (71.4%) were true-negatives. Only 1 of 15 (6.7%) predicted to deliver in less than 24 hours delivered in the ideal SADI (false-negative) in the 28 to 33 week group. Analysis of those predicted to deliver in greater than 7 days is not as fruitful since many were excluded for other reasons; nonetheless, 5 of 7 were correctly predicted.

Comment If one accepts that risk to benefit considerations are important in all decisions, then the provision of data which better defines and separates that segment of the population that will derive no benefit from that which will is clearly significant. Such data not only reduce the likelihood of future short- or long-term sequelae if the decision involves whether or not to administer a drug in pregnancy, but they also allow the physician to identify subsets within the population at risk, each with a different risk to benefit ratio.

for neonates 28-33

12 (1)

tModerate RDS. tlntervals of 3, 37, 27, and 24 days to 36, 33, 31, and 34 weeks. of 72, 29, and

with RDS)

TOtal

RDS.

SIntervals

(No.

13

*Moderate t Intervals

Table

(2)

345

No.

Anticipate Anticipate

< 24 hours > 7 days

Patient refusal Cervix 5 cm dilated Prior steroids L/S ratio Z 2 Note: *L/S

More ratio

weeks 5%

with RDS 34-37 No.

weeks %

15

75

i

71

5 6 1 1 1*

25 30 5 5 5

2 I 0 1) I)

29 14 0 0 0

than one exclusion criteria per 2 : 1 in presence of meconium.

patient.

In the only large prospective randomized study, Liggins has indicated the predictive significance of a SADI of 1 to 7 days as well as gestational age of less than 34 weeks, particularly at 30 and under 32 weeks.‘O Block subsequently emphasized the importance of surfactant assessment.16 However, the three variables have not been incorporated in a large series or evaluated prospectively. Previous studies also concentrated on the population treated with steroids and provided little or no data on the excluded population. As a result, it is not known what percentage of patients less than 37 weeks are truly candidates for st.eroids, what happens to those that are designated noncandidates. and how accurate are our designations in both the treated and untreated groups. In 439 patients screened to determine eligibility for steroid therapy, 392 (89.7%) were excluded. One hundred ninety-four of 229 patients (84.7%) between 28 to 33 weeks and 198 of 210 patients (94.3%) between 34 to 37 weeks were excluded. ‘Thus, a sig-

346

Depp

et al.

T&e X. Actual versus predicted screen to delivery interval in 27 patients excluded because of predicted SAD1 whose neonates developed RDS

~ <24 Hours 1-7 Days Total

14 I* 014

0

14

4

2

5

101 6

0

0

0

20

1

8 27

*Thirty-eight-hour interval. n&cant percentage of patients were not eligible using the two indicators of no benefit (incorrect interval and mature L/S ratio) plus other cIinicaI contraindi~tions mentioned in the material and methods section. Most patients were excluded because the predicted SADI was 1 to 7 days. This exclusion criteria occurred in 79% of 194 excluded patients at 28 to 33 weeks and in 87% of 198 patients excluded at 34 to 37 weeks. Concern that the SAD1 cannot be accurately predicted is realistic but not supported by our data. Of the 206 patients excluded because the predicted SAD1 was less than 24 hours, 85.4% delivered in that interval (truenegative). To reach the remaining 14.6% would have meant giving steroids to patients who would have received no benefit (false positive). Actually, only 12.6% delivered in the 2- to 7-day interval. Prediction of the ideal 2- to ?-day SADI was equally good. It was correctly predicted (true positive) in 61.2% of patients excluded for other reasons and in 66% of steroid therapy patients. Only 6 patients (12.8%) delivered in less than 24 hours (false positive). In summary, even though 114 patients (26.1%) were predicted to deliver in the 2 to 7 day interval, only 72 patients (63.2%) were correctly predicted (true positive). Unfortunately, 41 patients (56.9%) were excluded for other reasons. Finally, only 47 patients (10.9%) out of the entire series were ideal steroid candidates, Of these, only 3 1 patients (66%) or 6.9% of the total population delivered in the ideal SADI, thus representing the magnitude of the population which may benefit from steroids. If one applies the risk to benefit factors previously discussed to this 6.9%, it is clear that only a small percentage of patients at risk for delivery prior to 37 weeks can derive benefit; most are at less than 34 weeks. The impact of delaying delivery in the patients with premature labor, premature rupture of the membranes, or both, is largely one of added intrauterine maturation. Thirty-three of 103 patients (32%) and 8 of 42 patients (19%) with premature rupture of the

membranes and premature rupture of the membranes in the presence of premature labor were delayed greater than 48 hours. Twenty patients (19.4%) and 3 patients (7.1%) were delayed greater than 1 week. Inhibition of labor and avoidance of delivery prior IO established fetal pulmonary maturity are obviously keys in the prevention of RDS. Our data indicate that approximately 19% of all patients less than 34 weeks have mature lung function. Thirty-five percent of all specimens between 34 and 37 weeks were mature. Thus, amniocentesis should be performed not only to determine who should get steroids but also the significant population who do not require it. As a side benefit, if one obtains a very low L/S ratio, the effort to delay delivery is often intensified. L/S data was slightly different in the excluded patients because of selection. Twelve of 43 patients (48.8%) at 28 to 33 weeks had mature L/S ratios. Twenty-four of 61 patients (39.3%) at 34 to 37 weeks had mature L/S ratios. All neonates with RDS had immature L/S ratios with the exception of one patient at 28 to 33 weeks who had a ratio of 2 : 1 in the presence of meconium; 20 patients (10.3%) screened but excluded at 28 to 33 weeks, and 7 (3.5%) screened but excluded at 34 to 37 weeks, developed RDS. Twenty of the 27 neonates who developed RDS were excluded because of prediction of nonideal SADI. Concern that they were denied candidacy because of physician judgment error was not confirmed. Seventy percent in the 28 to 33 week and 71.4% in the 34 to 37 week group were true negatives and were appropriately excluded. Thus, it appears that even if the efficacy of steroids is substantiated by a large controlled study similar to that of Liggins most patients will have valid criteria which exclude steroid use. Further, it is possible to identify the population that is less likely to derive benefit. This group will continue to contribute a large number of RDS cases until the patient at risk for premature delivery can be identified earlier to reduce the number of patients currently included in the true negative group. To date, there has not been a single, large, well-controlled study to substantiate the findings of Liggins. For the sake of discussion, using the data of Liggins, it is possible to identify presently accepted factors which are useful in the selection of a population of patients in whom steroid administration may have potential benefit. A patient failing to fulfill the criteria of benefit is not likely to derive benefit. The most common contraindication for consideration of antenatal steroids in an anticipated delivery interval of less than 24 hours. Until early detection of the Iikelihood of

Volume Number

137 3

Antenatal

premature labor is possible, consideration of steroids in this group is probably unwarranted. Early and aggressive attempts to inhibit uterine activity are important techniques in reducing the incidence of prematurity and associated RDS. However, reduction of RDS and RDS morbidity will continue to depend upon an organized perinatal effort which includes: watchful ex-

115:29,

15.

16.

17.

1976.

7. Farrell, P. M., and Kotas, R.: Prevention of hyaline membrane disease: New concepts and approaches to therapy, Adv. Pediatr. 23:213, 1976. 8. Taeusch, J. W., Jr., and Avery, M. E.: Regulation of pulmonary alveolar development in late gestation and the prenatal period, in Hodson, A., editor: ~velopment of the Lung, New York, 1977, Marcel Dekker, Inc. 9. Liggins, G. C., and Howie, R. N.: A controlled trial of antepartum glucocorticoid treatment for prevention of the respiratory distress syndrome in premature infants, Pediatrics 50~5 15, 1972. 10. Liggins, G. C.: The Prevention of RDS by maternal betamethasone administration, in Lung Maturation and the Prevention of Hyaline Membrane Disease, Proceedings of the Seventieth Ross Conference on Pediatric Research, Columbus, Ohio, 1976, Ross Laboratories, p. 189. II. Fargier, P., Salle, B., Baud, N., et al.: prevention du syndrome de detresse respiratoire chez la premature, Nouv. Presse Med. 3:1595, 1974. 12 Caspi, E., Schreyer, P., Weinraub, Z., et al.: Prevention of the respiratory distress syndrome in premature infants by antepartum glucocorticoid therapy, Br. J. Obstet. Gynaecol. 83:187, 1976. 13. Dluholucky, S., Babic, J., and Taufer, I.: Reduction of incidence and mortality of respiratory distress syndrome by administ~tion of hydr~or~isone to mother. Arch. Dis. Child. 51:420, 1976. 14. Stocker, J.: Clinical trials with betamethasone, & Lung

Discussion Editors’ note: This manuscript

was revised

after these

discussions were presented. DR. &C-D P. BEEIBEL, Minnea~lis, Minnesota. In this paper, Dr. Depp and his associates raise the question of cost to benefit ratio, or how many patients are we wilIing to treat in order to help one. I agree with the authors that it is valid to examine the cost to benefit ratio. It is something we should do more

to prevent

pectancy of patients with premature rupture branes, aggressive intrapartal care of the fetus especially with regard to fetal distress, ognition and maternal transport of patients premature delivery to Level III centers, and pediatric assessment and therapy.

REFEREtQCES

1. Farrell, P. M., and Wood, R. M.: Epidemiolo~ of hyaline membrane disease: anafysis of national mortality statistics, Pediatrics 58:167, 1976. 2. Idiopathic respiratory distress syndrome, Proceedings of Interdisciplinary Conferences, U. S. Government Printing Office, Washington, D. C., 1969. 3.I Klauss, M.: Care of High Risk Neonate, Philadelphia, 1973, W. B. Saunders Co., p. 128. 4. Farrell. P. M.. and Averv. M. E.: Hvaline membrane disease, Am. Re;. Respir. & flf:65$, 1975. 5. Avery, M. E.: Phar~cologic approach to the acceleration of fetal tung maturation, Br. Med. BulI. 31:13, 1975. 6. Ballard, P. L., Benson, A., and Brehier, A.: Glucocorticoid effects in the fetal lung, Am. Rev. Respir. Dis.

corticosteroids

18.

19.

20.

21. 22. 23. 24. 25.

RDS

347

of mempremature early recat risk for aggressive

Maturation and the Prevention of Hyaline Membrane Disease, Proceedings of the Seventieth Ross Conference on Pediatric Research, Columbus, Ohio, 1976, Ross Laboratories, p. 2 12. Kennedy, J. L., Jr.: Clinical experience with betamethasone for the prevention of respiratory distress syndrome, in Lung Maturation and the Prevention of Hyaline Membrane Disease, Proceedings of the Seventieth Ross Conference on Pediatric Research, Columbus, Ohio, 1976, Ross Laboratories, p. 181. Block, M. F., KIing, 0. R., and Crosby, W. M.: Antenatal gluc~o~coid therapy for the prevention of respiratory distress syndrome in the premature infant, Obstet. Gynecol. 50: 186, 1977. Ballard, P. L., Carter, J., and Ballard, R. A.: prenatal betamethasone for the prevention of idiopathic respiratory distress syndrome, -Clin. Res. 25: 187.& 1977. Ballard. R. A.. Ballard. P. L.. Granbern. 1. P.. and Snider&an, S.: Prenatal admini&ation of b%ð&one for prevention of respiratory distress syndrome, J. Pediatr. 94~97, 1979. Taeusch, H. W., Jr., Frigoletto, F., Kitzmiller, J., Avery, M. E., Hehre, A., Fromm, B., Lawson, E., and Neff, R. K.: Risk of respiratory distress syndrome after prenatal dexamethasone treatment, Pediatrics 63:64, 1979. Papageorgiou, A. N., Desgranges, M. F., Masson, M., Colle, E., Shatz, R., and Gelfand, M. N.: The antenatal use of betamethasone in the prevention of respiratory distress syndrome: A controlled double-blind study, Pediatrics 63:73, 1979. Gluck, L., and Kulovich, M. V.: Le~~hin/sphingomyelin ratios in amniotic fluid in normal and abnormal pregnancy, AM. J. OBSTET. GYNECOL. 115:541, 1973. Gluck, L.: Administration of tort&steroids to induce maturation of fetal lung, Am. J. Dis. Child. 130~976, 1976. Ballard, R. A., and Ballard, P. L.: Use of prenatal glucocorticoid therapy to prevent RDS, Am. J. Dis. Child. 130:982, 1976. Taeusch, H. W., Jr.: Glucocorticoid prophylaxis for resplratory distress syndrome: A review of potential toxicity, J. Pediatr. 87~617, 1975. Litwack, G., Biochemical Actions of Hormones, New York, 1975, vol. 3, Academic Press Inc.

often in medicine. However, it is difficult to examine this ratio when we don’t know what the cost is. The real cost would be any possible side effects to mother and baby, especially to those who were treated and didn’t really need the steroids or were not helped by them. This is unknown and, therefore, makes it impossible to calculate the ratio. On the other hand, if there were no side effects, one might he willing to treat dozens of patients in order to help one.

348 Depp et al.

Also, the patient herself should participate in the decision as much as possible. Some patients would go with odds of 10 to 1, but not with 100 to 1. There is no discussion in the paper of the criteria used by the authors to predict when the patients would deliver. How did they decide whether a patient would deliver in the ideal time period or out of it? Let’s take two examples from the paper. The authors excluded 325 patients because they were predicted to deliver either before 24 hours or greater than 7 days after the time they were initially screened. Yet, 40 of the patients (12%) actually delivered in the ideal time as defined in the paper, that is, greater than 24 hours and less than 7 days after being considered for steroids. Should on treat seven patients in order to possibly benefit the eighth? Some physicians and patients would say yes. Second, of the excluded patients, 105 actually delivered greater than 7 days after the time steroids were considered. The vast majority of these patients should have had L/S ratios done and treatment should have been given to those with immature ratios. The authors do not tell us how many of these patients actually had an L/S ratio done. Including this group in the overall number does not seem valid. Nowhere in the paper do the authors tell us exactly how many excluded neonates developed RDS. They do tell us that of those excluded because they were predicted to deliver outside the ideal time bracket, 27 neonates developed RDS. Twenty delivered in less than 24 hours and, therefore, would not be expected to be helped by steroids. Seven might have been helped. Using this minimum of 7 out of 325, the ratio is approximately 1 out of 46. For comparison, when genetic amniocentesis is done for women in the age range of 35 to 39 years, a chromosomal abnormality is found only 1.5% of the time, or 1 out of 67. Until we know what the cost is, that is, the side effects of steroid administration, it is impossible to make a valid assessment of the cost to benefit ratio. DR. DWIGHT P. CRUIKSHANK, Iowa City, Iowa. Before commenting on specific aspects of this paper, it is essential to address the crucial question of whether, at this time, the use of corticosteroids to accelerate fetal lung maturation can be justified in any patient. This question has not been answered to date, nor can it be answered until the results of the second part of this study, and many more like it, become known. There is an increasingly large body of animal data which points to potentially serious long-term sequelae of such therapy, especially neurologic sequelae. Treatment of fetal guinea pigs with dexamethasone depresses incorporation of thymidine into DNA in the cerebrum, cerebellum, and medulla, and cortisol treatment of neonatal rats lead to a permanent deficit in brain cell number.’ Rats thus treated also show permanent alteration of circadian rhythms and desynchronization of the normal sequence of events at pu-

berty.’ Epstein and associates” treated pregnant rhesus monkeys with betamethasone at a dose per kilogram body weight and at a time in gestation very similar to the regimens now being used clinically. I-hirty-eight percent of the offspring of these monkeys had histologic changes in the brain suggesting neuronal injury, with shrunken nerve cells and gliosis. Granted, it is difficult to compare the effects of drugs given to different species at different stages of gestation, but general principles of embryonic and fetal growth apply to all animals: it is hazardous to completely ignore such animal data. Furthermore, the growth pattern of the human brain is very similar to that of the rat,’ with a myelinization period which persists long after birth. and with cell division and migration occurring in the cerebellum throughout the entire first year of life. Neurologic sequelae are especially serious because brain development follows a set pattern of cell multiplication, migration, and intersection; neurologic alterations in the fetus, unlike changes in other organs, probably cannot be overcome by a postnatal “catch-up” phenomenon. Other reported sequelae in animals treated with corticosteroids as fetuses or neonates include increased rates of fetal loss, diminished fetal weight, altered immune responses and accelerated maturation in liver. kidney, and adrenals. 3, 1 While reports of adverse sequelae in humans thus treated are rare, neonatal leukemoid reactions5 and altered leukocyte function” have been reported. Furthermore, no human neonates have been foIlowed long enough to know whether adverse sequelae will emerge. The history of perinatology is filled with therapeutic misadventures, the long-term effects of which took years to become apparent. These include use or misuse of sulfonamides, chloramphenicol, hexachlorophene, vitamin K, oxygen, and ofcourse diethylstilbestrol, the adverse sequelae of which did not become apparent for two to three decades. Nonetheless, as the authors point out, HMDlRDS is the major problem in perinatology, responsible for thousands of deaths and great morbidity each year; thus, an agent which promises to prevent it has great appeal to all obstetricians and pediatricians. Corticosteroid use is a fact of life, and thus, if it is going to be used, the issue of how it should be used must be addressed. As Depp and associates point out, the definitive study regarding the human use of corticosteroids remains that of Liggins and Howie,i other studies being smaller and much less well controlled. It is important to emphasize that Liggins and Howie could demonstrate no benefit for the use of glucocorticoids after 32 weeks’ gestation, in terms of reducing the incidence or mortality of HMD/RDS. Many infants beyond 32 weeks have mature L/S ratios; many of those who do not have mature lungs, do not develop HMD/RDS regardless. Liggins and Howie also demonstrated a worsening of perinatal outcome when corticosteroids were used in pre-eclamptic patients. One great hazard in the

Volume Number

157 3

widespread use of corticosEeroids is that we as obstetricians may become cavalier and complacent about premature delivery, feeling that the fetus is “protected.” I am aware of a physician who gives betamethasone to all patients prior to repeat cesarean section and elective induction. This is a deplorable and indefensible practice. Turning to the specifics of this paper, its most important message is that very few patients, 10.7%, with threatened premature delivery are candidates for corticosteroid therapy. If we consider both the excluded and treated group, only 26% of the patients delivered within the optimal steroid administration to delivery interval (SADI), namely greater than 24 hours and less than 7 days. With compulsive and expert evaluation, the authors were able to exclude many patients who would not deliver within the optimal interval, so that 66% of their treated group delivered after 24 hours but before 7 days. The most disturbing aspect of this, however. is that despite their best efforts, 34% of the patients treated with steroids delivered outside the optimal interval: thus, these fetuses received all of the theoretical hazards of the agent and none of the benefits. Another important fact revealed by this study is that 28% of the patients on whom amniocentesis was performed had mature L/S ratios, including 19Yo at 28 to 33 weeks. Since amniocentesis was performed on only 3 1% of the patients, it is entirely possible that the 28% figure is a low estimate. Nonetheless, in this group of patients with threatened premature delivery, at least 28% could not benefit from corticosteroids because of preexisting pulmonary maturity. When coupled with the 74% of the total group which delivered outside the optimal time interval, it is apparent, as the authors emphasize, that the population which theoretically can benefit from corticosteroid therapy is quite small. In summary, I do not believe antenatal corticosteroids should be used in humans to prevent RDS until further animal studies prove the safety of this approach. However, if they are to be used, it must be done with adherence to several rigid criteria. First, careful attention must be paid to accurate prediction of SADI, for as the authors have shown, this will increase the frequency of patients who might benefit from 26% to 66%. Second, steroids should not be used after 32 weeks’ gestation or in pre-eclamptic patients. Third, they should not be used in patients between 28 to 32 weeks unless amniocentesis reveals an immature L/S ratio, to avoid unnecessary risk to the 19% of this group the authors found to have mature lungs. Last, they must be used only under conditions of an experimental protocol, ideally randomized, certainly with written informed patient consent and a plan of longterm (20 to 30 years) follow-up of the offspring. I congratulate Dr. Depp and his colleagues on this paper. Certainly this approach to the problem is unique and important, and it proves that the popula-

Antenatal

corticosteroids

to prevent

RDS

349

tion which can benefit from these agents IS small. Whether the cost outweighs the benefit, however, will only be answered by the second half of their study, including long-term follow-up. I would like to ask Dr. Depp what the plan is for long-term follow-up of these infants. REFERENCES

Sanfacon, R., Possmayer, F., and Harding, P. G. R.: Dexamethasone treatment of the guinea pig fetus: Its effects on the incorporation of 3H-thymidine into deoxyribonucleic acid, AM. J. OBSTJZT. GYNEGOL. 127:745, 1977. 2. Weichsel, M. E.: The therapeutic use of glucocorticoid hormones in the perinatal period: Potential neurological hazards, Ann. Neurol. 4:364, 1977. 3. Epstein, M. F., Farrell, P. M., Sparks, J. N., Pepe. I;., Dris~011, S. G., and Chez, R. A.: Maternal betamethasone and fetal growth and development in the monkry. AM. J. 1.

OBSTET.

GYNECOL.

127:261,

1977.

Gluck, L.: Administration of corticosteroids to induce maturation of fetal lung, Am. J. Dis. Child. 180:976, 1976. 5. Bielowski, D., Hiatt, I. M., and He@, T.: Betamethasoneinduced Ieukaemoid reaction in pre-term infant, Lancet

4.

1:218, 6.

1978.

Lazzarin, A., Capsoni, F., Moroni, M., Pardi, G., and Marini, A.: Leucocyte function after antenatal betamethasone given to prevent respiratory distress, Lancet 2: 1354, 1977.

7. Liggins,

G. C., and

antepartum

Howie,

glucocorticoid

respiratory distress syndrome atrics 50:515, 1972.

R.

N.:

A controlled

treatment for prevention in premature

infants,

trial

of

of the Pedi-

DR. FEDERICO MARIONA, Detroit, Michigan. I think your numbers are very provocative, Dr. Depp. Faced with about 530 such patients a year under similar circumstances as you described, I have the exact same feeling. Our present rate of neonatal morbidity on fetuses that have received steroids is appr,oximately 38%.

Eleven out of 13 mothers that have been treated with steroids prenatally developed postpartum morbidity. The accuracy of the present neonatal follow-up is far from being desirable in terms of applying a potential deleterious agent. The study in puppies done in our own institution showed a 17% decrease in DNA in the brain of those puppies after being treated with comparative doses of steroid during intrauterine life. DR. EMANUEL GAZIANO, Minneapolis, Minnesota. Were any attempts made to inhibit labor in those patients who were anticipated to deliver within 24 hours, particularly in the low gestational age group? DR. DEPP (Closing). The prediction of delivery interval is a difficult task at best. We do not have a score. I am sorry that we did not develop one. However. factors that we used to predict the interval included cervical dilatation, the presence of cervical retraction on pelvic examination during the contraction phase, contraction frequency on an external monitor, and pelvic pressure with the contractions. We have observed that the presence of premature rupture of the membranes, with some bleeding, almost uniformly predicts the delivery within 24 hours.

350

Depp et al.

Forty-three of our 392 patients who were excluded developed RDS. Nineteen of the 20 cases in which the predicted interval was a factor were excluded for valid interval discrepancies. The remaining 23 cases were excluded for other valid reasons. Dr. Cruikshank really answered some of Dr. Bendel’s earlier remarks about the potential cost of steroids. Certainly we can learn enough from history not to have to relive it. We will continue to live with the diethylstilbestrol consequences for some time. The literature dealing with animals who have been administered steroids is large and involves almost all organ systems, including the lung, brain, and organs derived from the foregut. The point 1 believe is that steroids cannot, as

they presently are nationally, be given on almost ,111 indiscriminate manner to patients with poorly documented dates, uncertain surfactant status, and witlr little consideration for criteria of benefit vs. exclusion. Regarding follow-up, we anticipate following these patients for at least 5 years. If I had my way, 1 would like to do as Dr. Cruikshank indicates, and be in a position to follow them for the next 21 or more years. All patients except those wih amnionitis, who were at risk for early delivery, were aggressively treated with labor-inhibiting drugs as soon after admission as possible. Some maternal transport cases received medication enroute from outlying hospitals.