Aspirin For Evidence-Based Preeclampsia Prevention Trial: Effect of Aspirin On Length of Stay in The Neonatal Intensive Care Unit
Aspirin For Evidence-Based Preeclampsia Prevention Trial: Effect of Aspirin On Length of Stay in The Neonatal Intensive Care Unit
Aspirin For Evidence-Based Preeclampsia Prevention Trial: Effect of Aspirin On Length of Stay in The Neonatal Intensive Care Unit
org
OBSTETRICS
Aspirin for Evidence-Based Preeclampsia Prevention
trial: effect of aspirin on length of stay in the neonatal
intensive care unit
David Wright, PhD; Daniel L. Rolnik, MD; Argyro Syngelaki, PhD; Catalina de Paco Matallana, MD; Mirian Machuca, MD;
Mercedes de Alvarado, MD; Sofia Mastrodima, MD; Min Yi Tan, MD; Siobhan Shearing, RM; Nicola Persico, MD;
Jacques C. Jani, MD; Walter Plasencia, MD; George Papaioannou, MD; Francisca S. Molina, MD; Liona C. Poon, MD;
Kypros H. Nicolaides, MD
BACKGROUND: Preeclampsia is a major pregnancy complication with RESULTS: In the trial there were 1620 participants and 1571 neonates
adverse short- and long-term implications for both the mother and baby. were liveborn. The total length of stay in neonatal intensive care was
Screening for preeclampsia at 11e13 weeks’ gestation by a combination substantially longer in the placebo than aspirin group (1696 vs 531 days).
of maternal demographic characteristics and medical history with mea- This is a reflection of significantly shorter mean lengths of stay in babies
surements of biomarkers can identify about 75% of women who develop admitted to the neonatal intensive care unit from the aspirin than the
preterm preeclampsia with delivery at <37 weeks’ gestation and 90% of placebo group (11.1 vs 31.4 days), a reduction of 20.3 days (95% con-
those with early preeclampsia at <32 weeks, at a screen-positive rate of fidence interval, 7.0e38.6; P ¼ .008). Neonatal intensive care of babies
10%. A recent trial (Combined Multimarker Screening and Randomized born at <32 weeks’ gestation contributed 1856 (83.3%) of the total of
Patient Treatment with Aspirin for Evidence-Based Preeclampsia Pre- 2227 days in intensive care across both treatment arms. These occurred
vention) has reported that in women identified by first-trimester screening in 9 (1.2%) of the 777 livebirths in the aspirin group and in 23 (2.9%) of
as being at high risk for preeclampsia, use of aspirin (150 mg/d from the 794 in the placebo group (odds ratio, 0.42; 95% confidence interval,
first to the third trimester), compared to placebo, reduced the incidence of 0.19e0.93; P ¼ .033). Overall, in the whole population, including
preterm preeclampsia, which was the primary outcome, by 62% (95% 0 lengths of stay for those not admitted to the neonatal intensive care unit,
confidence interval, 26e80%) and the incidence of early preeclampsia by the mean length of stay was longer in the placebo than aspirin group (2.06
89% (95% confidence interval, 53e97%). The surprising finding of the vs 0.66 days; reduction of 1.4 days; 95% confidence interval, 0.45e2.81;
trial was that despite the reduction in preeclampsia the incidence of P ¼ .014). This corresponds to a reduction in length of stay of 68% (95%
admission to the neonatal intensive care unit, which was one of the confidence interval, 20e86%).
secondary outcomes, was not significantly affected (odds ratio, 0.93; 95% CONCLUSION: In pregnancies at high risk of preeclampsia adminis-
confidence interval, 0.62e1.40). tration of aspirin reduces the length of stay in the neonatal intensive care unit
OBJECTIVE: We sought to examine the effect of prophylactic use of by about 70%. This reduction could essentially be attributed to a decrease in
aspirin during pregnancy in women at high risk of preeclampsia on length the rate of births at <32 weeks’ gestation, mainly because of prevention of
of stay in the neonatal intensive care unit. early preeclampsia. The findings have implications for both short- and long-
STUDY DESIGN: This was a secondary analysis of data from the term health care costs as well as infant survival and handicap.
Aspirin for Evidence-Based Preeclampsia Prevention trial to assess evi-
dence of differences in the effect of aspirin on length of stay in neonatal Key words: aspirin, Combined Multimarker Screening and Random-
intensive care. Bootstrapping was used for the comparison of mean length ized Patient Treatment with Aspirin for Evidence-Based Preeclampsia
of stay between the aspirin and placebo groups. Logistic regression was Prevention trial, first-trimester screening, health economics, neonatal
used to assess treatment effects on stay in the neonatal intensive care unit. intensive care, preeclampsia
The statistical software package R Babies born at <32 weeks’ gestation Impact on cost
was used for data analyses.25 The R contributed to 1856 (83.3%) of the total In a population of 10,000 pregnancies
packages lme426 and boot27 were used of 2227 days in NICU across both undergoing first-trimester screening for
for mixed effects logistic regression and treatment arms; these occurred in 1.2% PE, at a screen-positive rate of 10%, 1000
bootstrapping. of livebirths in the aspirin group and in pregnancies would be classified as high
2.9% in the placebo group (odds ratio, risk. If these 1000 pregnancies had not
Results 0.42; 95% CI, 0.19e0.93). Admission to received aspirin and the mean length of
In the ASPRE trial there were 822 par- NICU occurred in all 32 babies born at stay in NICU was 2.06 days, the expected
ticipants in the placebo group and 798 in <32 weeks’ gestation, in 23 (67.6%) of total length of stay would be 2060 days. If
the aspirin group.21 There were no sig- 34 born at 32e34 weeks, in 13 (12.5%) they had received aspirin the expected
nificant differences between the aspirin of 104 born at 35e36 weeks, and in 34 total length of stay would be 660 days. It
and placebo groups in baseline charac- (2.4%) of the 1401 born at 37-42 weeks. is difficult to attach specific costs to daily
teristics.21 In the placebo group, there Prolonged stay in NICU for 14 days lengths of stay but, if we assume $4000,
were 16 miscarriages or pregnancy ter- contributed to 1914 (85.9%) of the total of then the cost saving from such care by a
minations at 24 weeks’ gestation, 12 2227 days across both treatment arms; policy of screening 10,000 pregnancies
stillbirths at 24 weeks, and 794 live- these occurred in 1.0% of livebirths in the and treating the high-risk group with
births. In the aspirin group, there were aspirin group and in 3.0% in the placebo aspirin would be $4000 (2060 e
14 miscarriages or pregnancy termina- group (odds ratio, 0.30; 95% CI, 660) ¼ $5.6 million. This is equivalent to
tions, 7 stillbirths, and 777 livebirths. 0.11e0.81) and this is a reflection of the $560 per patient screened, which is well
Rates of admission to NICU and reduction in the number of babies born at in excess of the cost of screening.
length of stay by treatment group are <32 weeks’ gestation. The length of stay in
shown in the Figure and Table. There NICU varied for individual babies from 1- Comment
was no significant difference in rates of 230 days; it was 1-3 days in 39 (38.2%) of Principal findings of this study
admission to NICU between the aspirin the 102 babies, 4-6 days in 11 (10.8%), 7 The ASPRE trial demonstrated that, in
and placebo groups (6.2% vs 6.8%; days in 52 (51.0%), 14 days in 32 women with singleton pregnancies
odds ratio, 0.94; 95% CI, 0.63e1.42). (31.4%), and 21 in 23 (22.5%). identified by means of first-trimester
However, the total length of stay in The effect of aspirin in reducing the screening as being at high risk for PE,
NICU in the aspirin group was sub- length of stay in NICU was partly the prophylactic use of aspirin reduces
stantially shorter than in the placebo mediated by a reduction in the rate of PE the incidence of preterm PE and early PE
group (531 vs 1696 days). This is a (Table). The incidence of babies by approximately 60% and 90%,
reflection of significantly shorter mean admitted to the NICU after delivery respectively.21 This secondary analysis
lengths of stay in babies admitted to because of PE was 2.3% (18 of 794) in the demonstrated that use of aspirin reduces
NICU from the aspirin group than placebo group and 0.3% (2 of 777) in the the length of stay in NICU by approxi-
from the placebo group (11.1 vs 31.4 aspirin group (odds ratio, 0.11; 95% CI, mately 70%. This reduction could
days), a reduction of 20.3 (95% CI, 0.02e0.50). In the pregnancies deliv- essentially be attributed to a decrease in
7.0e38.6) days (P ¼ .008). In the whole ering at <37 and <32 weeks’ gestation the rate of births at <32 weeks’ gestation,
population, including 0 values for those the admission to NICU was 2.0% and mainly because of prevention of early PE.
not admitted to the NICU, the mean 0.9%, respectively, in the placebo group The consequence of reduction in
length of stay was 2.06 days in the and 0.1% and 0% in the aspirin group length of stay in NICU is substantial
placebo group and 0.66 days in the (odds ratio, 0.06; 95% CI, 0.01e0.50 and saving in health care cost which is well in
aspirin group; therefore, aspirin odds ratio, 0.00; 95% CI, 0.00e0.56, excess of the cost of population screening
reduced the mean length of stay by an respectively). Aspirin also had a nonsig- and treatment of the high-risk group
estimated 1.40 days (95% CI, nificant effect in reducing the length of with aspirin. The study provides the
0.45e2.81 days; P ¼ .014). stay in NICU in pregnancies without PE basis for formal health economic studies.
The reduction in total length of stay in with delivery at <32 weeks’ gestation
NICU in the aspirin group could essen- (odds ratio, 0.59; 95% CI, 0.26e1.36). Strengths and limitations of this
tially be attributed to a decrease in the There were 16 (2.0%) babies from the study
rate of births at <32 weeks’ gestation, placebo group (8 after spontaneous ASPRE was a large multicenter trial that
mainly because of prevention of early birth, 7 after iatrogenic birth for fetal was powered for the primary outcome of
PE, and consequent decrease in number growth restriction, and 1 for maternal preterm PE and the statistical power for
of babies with prolonged stays of 14 indications) and 9 (1.2%) from the detecting less frequent outcomes is inev-
days (Figure and Table). After 32 weeks’ aspirin group (6 after spontaneous birth, itably poor. This secondary analysis was
gestation there is flattening in the cu- 2 after iatrogenic birth for fetal growth triggered by the apparent contradiction
mulative length of stay in NICU for both restriction, and 1 for maternal that although aspirin use was associated
the aspirin and placebo groups (Figure). indications). with a major reduction in preterm and
FIGURE
Admission to neonatal intensive care unit in the trial groups
55 1700 55
1600
50
1500
45 1400 45
15 500
15
400
10 300 10
200
5 5
100
0 0 0
24 26 28 30 32 34 36 38 40 42 24 26 28 30 32 34 36 38 40 42 24 26 28 30 32 34 36 38 40 42
Gestation at birth (w) Gestation at birth (w) Gestation at birth (w)
Cumulative number of babies admitted to neonatal intensive care unit (NICU) according to gestational age at birth for placebo (blue circles) and aspirin
(red circles) groups. Cumulative NICU: number of all babies admitted (left), length of stay (center), and number of babies with length of stay >14 days.
Wright et al. Secondary analysis of ASPRE trial. Am J Obstet Gynecol 2018.
early PE there was no evidence of reduc- 450e2810 days. Assuming a cost of Our approach to screening is to use
tion in NICU admission. In this respect, $4000 per day, the corresponding in- Bayes theorem to combine information
the findings that first, babies born at <32 tervals for the cost saving would range from maternal factors with biophysical
weeks’ gestation contributed >80% of from $1.8-11.2 million, which equate to and biochemical measurements obtained
total length of stay in NICU; second, the between $180-1120 per screening test. at 11e13 weeks’ gestation to derive the
incidence of birth at <32 weeks was lower patient-specific risk. The method, which
in the aspirin group; and third, in the Prediction and prevention of PE detects around 75% of cases of preterm
aspirin group the total length of stay in The traditional approach of identifying PE at false positive rate of 10%, was
NICU was substantially reduced, are not women at high risk of PE who could originally developed from a study of
surprising. potentially benefit from the prophylactic 58,884 pregnancies,13,14 updated with
However, it has to be recognized that use of aspirin is based on maternal data from prospective screening in 35,948
this is an unplanned secondary analysis characteristics and features of the med- pregnancies,15,16 and subsequently vali-
and, because of the small number babies ical and obstetrical histories.28,29 How- dated in 2 independent data sets derived
with longer lengths of stay, there is ever, the performance of such screening from multicenter studies in 8775 and
considerable uncertainty in the estima- is poor. With the method recommended 25,797 pregnancies, respectively.32,33
tion of the difference in mean length of by the National Institute for Health and Prophylactic use of aspirin was previ-
stay between the aspirin and placebo Care Excellence in the United Kingdom, ously thought to reduce the risk of PE by
groups. Including 0 lengths of stay for the detection rate of preterm PE is about only 10%.17 However, recent evidence
those not admitted to the NICU, the 40% at screen-positive rate of 10% and suggests that the target for first-trimester
95% CI for the difference in mean length with the method recommended by the screening should be severe PE leading to
of stay ranged from 0.45e2.81 days. In a American Congress of Obstetricians and preterm birth, rather than all PE. In
screened population of 10,000 pregnan- Gynecologists in the United States the ASPRE, use of aspirin was associated
cies, treating 10% screened positive, this detection rate is 90% but at a screen- with a 62% reduction in the rate of
CI translates into an interval from positive rate of 67%.28-31 preterm PE with no significant effect on
2. Mayama M, Uno K, Tano S, et al. Incidence of at 11-13 weeks’ gestation. Am J Obstet Gyne- guidelines and ACOG recommendations. Ultra-
posterior reversible encephalopathy syndrome col 2016;214:103.e1-12. sound Obstet Gynecol 2017;49:756-60.
in eclamptic and patients with preeclampsia with 17. Askie LM, Duley L, Henderson-Smart DJ, 32. O’Gorman N, Wright D, Poon LC, et al.
neurologic symptoms. Am J Obstet Gynecol Stewart LA. Antiplatelet agents for prevention of Accuracy of competing-risks model in
2016;215:239.e1-5. pre-eclampsia: a meta-analysis of individual screening for pre-eclampsia by maternal fac-
3. Palatnik A, Grobman WA, Miller ES. Is a his- patient data. Lancet 2007;369:1791-8. tors and biomarkers at 11-13 weeks’ gesta-
tory of preeclampsia associated with an 18. Meher S, Duley L, Hunter K, Askie L. Anti- tion. Ultrasound Obstet Gynecol 2017;49:
increased risk of a small for gestational age infant platelet therapy before or after 16 weeks’ 751-5.
in a future pregnancy? Am J Obstet Gynecol gestation for preventing preeclampsia: an indi- 33. Rolnik DL, Wright D, Poon LC, et al. ASPRE
2016;215:355.e1-6. vidual participant data meta-analysis. Am J trial: performance of screening for preterm
4. White WM, Mielke MM, Araoz PA, et al. Obstet Gynecol 2017;216:121-8.e2. pre-eclampsia. Ultrasound Obstet Gynecol
A history of preeclampsia is associated with a 19. Roberge S, Nicolaides K, Demers S, Hyett J, 2017;50:492-5.
risk for coronary artery calcification 3 decades Chaillet N, Bujold E. The role of aspirin dose on 34. Wright D, Poon LC, Rolnik DL, et al. Aspirin
later. Am J Obstet Gynecol 2016;214:519.e1-8. the prevention of preeclampsia and fetal growth for evidence-based preeclampsia prevention
5. Fields JA, Garovic VD, Mielke MM, et al. restriction: systematic review and meta-analysis. trial: influence of compliance on beneficial ef-
Preeclampsia and cognitive impairment later in Am J Obstet Gynecol 2017;216:110-20.e6. fect of aspirin in prevention of preterm pre-
life. Am J Obstet Gynecol 2017;217:74.e1-11. 20. Roberge S, Bujold E, Nicolaides KH. Aspirin eclampsia. Am J Obstet Gynecol 2017;217:
6. Auger N, Rhéaume MA, Paradis G, Healy- for the prevention of preterm and term pre- 685.e1-5.
Profitós J, Hsieh A, Fraser WD. Preeclampsia eclampsia: systematic review and metaanalysis. 35. Pierrat V, Marchand-Martin L, Arnaud C,
and the risk of cataract extraction in life. Am J Am J Obstet Gynecol 2018;218:287-93.e1. et al. Neurodevelopmental outcome at 2 years
Obstet Gynecol 2017;216:417.e1-8. 21. Rolnik DL, Wright D, Poon LC, et al. Aspirin for preterm children born at 22 to 34 weeks’
7. Kajantie E, Osmond C, Eriksson JG. Gesta- versus placebo in pregnancies at high risk for gestation in France in 2011: EPIPAGE-2 cohort
tional hypertension is associated with increased preterm preeclampsia. N Engl J Med 2017;377: study. BMJ 2017;358:j3448.
risk of type 2 diabetes in adult offspring: the 613-22. 36. Larroque B, Ancel PY, Marret S, et al. Neu-
Helsinki birth cohort study. Am J Obstet Gynecol 22. Poon LC, Zymeri NA, Zamprakou A, rodevelopmental disabilities and special care of
2017;216:281.e1-7. Syngelaki A, Nicolaides KH. Protocol for mea- 5-year-old children born before 33 weeks of
8. Saleem S, McClure EM, Goudar SS, et al. and surement of mean arterial pressure at 11-13 gestation (the EPIPAGE study): a longitudinal
the Global Network Maternal Newborn Health weeks’ gestation. Fetal Diagn Ther 2012;31: cohort study. Lancet 2008;371:813-20.
Registry Study Investigators. A prospective 42-8. 37. Moster D, Lie RT, Markestad T. Long-term
study of maternal, fetal and neonatal deaths in 23. Plasencia W, Maiz N, Bonino S, Kaihura C, medical and social consequences of preterm
low- and middle-income countries. Bull World Nicolaides KH. Uterine artery Doppler at 11þ0 to birth. N Engl J Med 2008;359:262-73.
Health Organ 2014;92:605-12. 13þ6 weeks in the prediction of pre-eclampsia.
9. Lisonkova S, Joseph KS. Incidence of pre- Ultrasound Obstet Gynecol 2007;30:742-9.
24. Brown MA, Lindheimer MD, de Swiet M, Van Author and article information
eclampsia: risk factors and outcomes associ-
Assche A, Moutquin JM. The classification and From the University of Exeter, Exeter, United Kingdom
ated with early- versus late-onset disease. Am J
Obstet Gynecol 2013;209:544.e1-12. diagnosis of the hypertensive disorders of (Dr Wright); King’s College Hospital, London, United
pregnancy: statement from the International Kingdom (Drs Rolnik, Syngelaki, Machuca, de Alvarado,
10. Irgens HU, Reisaeter L, Irgens LM, et al.
Society for the Study of Hypertension in Preg- Mastrodima, Tan, and Nicolaides); Hospital Clı́nico Uni-
Long term mortality of mothers and fathers after
nancy (ISSHP). Hypertens Pregnancy 2001;20: versitario Virgen de la Arrixaca, Murcia, Spain (Dr Paco
pre-eclampsia: population based cohort study.
IX-XIV. Matallana); Medway Maritime Hospital, Gillingham,
BMJ 2001;323:1213-7.
25. R Development Core Team. R. A language United Kingdom (Dr Machuca); Homerton University
11. Li R, Tsigas EZ, Callaghan WM. Health and
and environment for statistical computing. Hospital, London, United Kingdom (Dr de Alvarado); North
economic burden of preeclampsia: no time for Vienna (Austria): R Foundation for Statistical Middlesex University Hospital, London, United Kingdom
complacency. Am J Obstet Gynecol 2017;217: Computing; 2011. (Dr Mastrodima); Lewisham University Hospital, London,
235-6.
26. Bates D, Maechler M, Bolker B, Walker S. United Kingdom (Dr Tan); Southend University Hospital,
12. Stevens W, Shih T, Incerti D, et al. Short- Fitting linear mixed-effects models using lme4. Essex, United Kingdom (Ms Shearing); Ospedale Mag-
term costs of preeclampsia to the United States J Stat Softw 2015;67:1-48. giore Policlinico, Milan, Italy (Dr Persico); University
health care system. Am J Obstet Gynecol 27. Canty A, Ripley B. Boot: bootstrap R (S-Plus) Hospital Brugmann, Université Libre de Bruxelles, Brus-
2017;217:237-48.
functions. R package 2017;1:3-20. sels, Belgium (Dr Jani); Hospiten Group, Tenerife, Canary
13. Wright D, Akolekar R, Syngelaki A, Poon L, 28. National Collaborating Center for Women’s Islands, Spain (Dr Plasencia); Attikon University Hospital,
Nicolaides KH. A competing risks model in early and Children’s Health (UK). Hypertension in Athens, Greece (Dr Papaioannou); Hospital Universitario
screening for preeclampsia. Fetal Diagn Ther pregnancy: the management of hypertensive San Cecilio, Granada, Spain (Dr Molina); and Kings’
2012;32:171-8. disorders during pregnancy. London: RCOG College, London, United Kingdom (Dr Poon).
14. Akolekar R, Syngelaki A, Poon L, Wright D, Press; 2010. Received Jan. 2, 2018; revised Feb. 14, 2018;
Nicolaides KH. Competing risks model in early 29. ACOG. First-trimester risk assessment for accepted Feb. 26, 2018.
screening for preeclampsia by biophysical and early-onset preeclampsia. Committee opinion Supported by grants from the Fetal Medicine
biochemical markers. Fetal Diagn Ther 2013;33: no. 638. Obstet Gynecol 2015;126:e25-7. Foundation (charity no. 1037116) and by the European
8-15. 30. Gallo DM, Wright D, Casanova C, Union 7th Framework ProgrameFP7-HEALTH-2013-
15. Wright D, Syngelaki A, Akolekar R, Poon L, Campanero M, Nicolaides KH. Competing risks INNOVATION-2 (Aspirin for Evidence-Based Preeclamp-
Nicolaides KH. Competing risks model in model in screening for preeclampsia by maternal sia Prevention project no. 601852). These bodies had no
screening for preeclampsia by maternal char- factors and biomarkers at 19-24 weeks’ gesta- involvement in the study design; in the collection, anal-
acteristics and medical history. Am J Obstet tion. Am J Obstet Gynecol 2016;214:619.e1-17. ysis, and interpretation of data; in the writing of the report;
Gynecol 2015;213:62.e1-10. 31. O’ Gorman N, Wright D, Poon LC, et al. or in the decision to submit the article for publication.
16. O’Gorman N, Wright D, Syngelaki A, et al. Multicenter screening for preeclampsia by The authors report no conflict of interest.
Competing risks model in screening for pre- maternal factors and biomarkers at 11-13 Corresponding author: Kypros H. Nicolaides, MD.
eclampsia by maternal factors and biomarkers weeks’ gestation: comparison to NICE kypros@fetalmedicine.com