| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ORIGINAL RESEARCH |
From the 1Department of Obstetrics and Gynaecology, Cambridge University, Cambridge, United Kingdom; and 2Harris Birthright Research Centre for Fetal Medicine, King’s College Hospital Medical School, London, United Kingdom.
 |
ABSTRACT |
|---|
TOP
 ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
METHODS: Data were available from 30,519 unselected women from seven units in the UK who had uterine artery Doppler performed between 22 and 24 weeks of gestation. The risk of stillbirth (n=109) was assessed using time to event and logistic regression analysis. Stillbirths were subdivided into placental (due to abruption, preeclampsia, or growth restriction) or unexplained.
RESULTS: The risk of placental stillbirth was increased among women with a mean pulsatility index in the top decile (adjusted hazard ratio [HR] 5.5, 95% confidence interval [CI] 2.8–10.6) and those with a bilateral notch (adjusted HR 3.9, 95% CI 2.0–7.8). The relationship between a mean pulsatility index in the top decile and the risk of unexplained stillbirth was weaker (adjusted HR 2.5, 95% CI 1.1–5.6) and there was no association with a bilateral notch. Placental stillbirths occurred at earlier gestations than unexplained stillbirths (median [interquartile range] 30 [26–36] compared with 38 [36–40], P<.001). Consequently, being in the top 5% of predicted risk of stillbirth on the basis of the combination of mean pulsatility index and notching was a good predictor (sensitivity, specificity, and positive likelihood ratio) of all cause stillbirth up to 32 weeks (58%, 95%, and 12.1, respectively) but a poor predictor of stillbirth at later gestations (7%, 95%, and 1.3, respectively).
CONCLUSION: Abnormal uterine artery Doppler was a better predictor of the risk of stillbirth due to placental causes than unexplained stillbirth. Consequently, abnormal uterine artery Doppler was a good predictor of stillbirth at extreme preterm gestations but a poor predictor of stillbirth at term.
LEVEL OF EVIDENCE: II
|
MATERIALS AND METHODS |
|---|
|
TOP
ABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The current analysis used this cohort but excluded women recruited to the trial of cervical cerclage. Maternal details and past medical and obstetric history were obtained using a questionnaire. Uterine artery Doppler studies were performed by using transvaginal ultrasonography10 by ultrasonographers trained in this method. Each uterine artery was identified using color flow mapping, and three similar consecutive waveforms were obtained using pulsed wave Doppler. The pulsatility index was measured, and the mean pulsatility index of the two uterine arteries was calculated.6 Quality control of screening, handling of data, and verification of adherence to protocols at the different centers was performed on a regular basis by the trial coordinators. The study period was October 1999 to August 2002. Women with a mean pulsatility index greater than 1.6, which in an earlier study represented the 95th centile,10 were followed up with growth scans, blood pressure measurements, and urinalysis for protein at 28, 32, and 36 weeks. Women with normal uterine artery Doppler received routine antenatal care. Maternal history and Doppler findings were recorded in a computer database at the time of the Doppler studies in each participating center.
For the purposes of dichotomizing uterine mean pulsatility index, those women with values in the top 10% were considered to have elevated mean pulsatility index; this was 1.43 and above in the present analysis. Outcome was ascertained by computerized databases in each of the centers. In all cases of adverse outcome (abruption, preterm birth, preeclampsia, and stillbirth), the clinical case record (and stillbirth autopsy, where performed) was reviewed by a medically qualified individual, the diagnosis confirmed, and further details were obtained, as necessary. Stillbirths were defined as delivery of an infant which showed no signs of life. Losses due to congenital abnormality were excluded. Stillbirth was defined as all intrauterine fetal deaths subsequent to the measurement of uterine artery Doppler. The median gestational age of assessment of uterine artery Doppler was 23 weeks. Stillbirths were divided into those where the fetus was thought to have died before the onset of labor (antepartum) and those where the fetus was thought to have been alive at the start of labor. All analyses for the present study focused on antepartum stillbirths. The presumed cause of stillbirth was obtained from the case notes. Where the women suffered severe preeclampsia or had an abruption, these were assumed to have caused the stillbirth. Abruption was diagnosed on the basis of clinical presentation and the presence of a retroplacental clot. Preeclampsia was defined according to the guidelines of the International Society for the Study of Hypertension in Pregnancy. This requires two recordings of diastolic blood pressure of more than 90 mm Hg at least 4 hours apart in previously normotensive women, and proteinuria of 300 mg or more in 24 hours, or two readings of at least two pluses on dipstick analysis of midstream or catheter urine specimens if no 24-hour collection is available.
In the absence of a direct cause, stillbirths were regarded as unexplained. These were subdivided into those where the birth weight was less than the 10th percentile for sex and gestation (small for gestational age) and those where the birth weight percentile was above this threshold (appropriate for gestational age). Unexplained stillbirths where the fetus was small for gestational age were assumed to be due to intrauterine growth restriction. Stillbirths due to preeclampsia, abruption, or small for gestational age unexplained stillbirths were collectively considered to be placentally related stillbirths.
Body mass index was defined as the woman’s prepregnancy weight in kilograms divided by her height in meters squared. Primigravid women were defined as those who had no previous births and any previous pregnancies had been spontaneous losses before 16 weeks or therapeutic abortions. Previous preterm birth was defined as a woman who had any previous live birth before 37 weeks of gestation. Previous loss was defined as any woman who had given birth to an infant showing no signs of life at or after 16 weeks of gestation and thus included prior intrauterine fetal deaths and stillbirths. Parous women with no previous complications were defined as those whose previous births had all been live births at term.
Continuous variables were summarized by the median and interquartile range (IQR) and comparisons between groups were made using the Mann-Whitney U test. Univariable comparisons of dichotomous data were made using the 
2 test. The P values for all hypothesis tests were two-tailed, and statistical significance was set at P<.05. The risk of stillbirth was compared between groups using time-to-event analyses (Kaplan-Meier and Cox proportional hazards model) in which gestational age was used as the time scale, antepartum stillbirth due to the specified cause was defined as the event, and all other births were treated as censored. This method uses ongoing pregnancies as the denominator, as previously suggested,11 but accounts for censoring due to birth, allows multivariable analysis,12 and can be used in situations where not all individuals would ultimately experience the event.13 This analytic approach allows assessment of the relative risk accounting for variation in the duration of pregnancy. Survival data were plotted as cumulative percentage, with event as recommended for rare outcomes,14 and univariable statistical comparisons were made using the log rank test. Crude and adjusted hazard ratios were estimated using the proportional hazards model.15 The proportional hazards assumption was tested using the test of Grambsch and Therneau16 as previously described for the analysis of stillbirth risk.17,18 Logistic regression analysis was used to estimate adjusted odds ratios within given gestational windows. In these analyses, the number of antepartum stillbirths within the given range was the numerator and the number of all births at the given or later gestations was the denominator. Mean pulsatility index was treated as a continuous variable in the logistic regression analysis. Linearity in the log odds scale was assessed using fractional polynomials.19 For predictive models, variables were selected using backward stepwise logistic regression,20 with the threshold for removal being P=.05. Logistic regression models were converted to tables of likelihood ratios using a previously described method.21 All statistical analyses were performed using the Stata 8 software package (Stata Corporation, College Station, TX).
|
RESULTS |
|---|
|
TOP
ABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
A uterine artery mean pulsatility index in the top decile and a bilateral notch were both associated with an increased risk of all-cause stillbirth and stillbirth due to a placental cause (Table 2). A uterine artery mean pulsatility index in the top decile was also associated with the risk of unexplained stillbirth. There were no associations between a unilateral notch and the risk of any type of stillbirth. The relationship between a composite index of abnormal Doppler (uterine mean pulsatility index in the top decile or bilateral notch or both) and the Kaplan-Meier cumulative risk of stillbirth is plotted in Figure 2. This illustrates that abnormal uterine artery Doppler was more strongly associated with all-cause stillbirth and stillbirth due to placental causes at extreme preterm gestations (Fig. 2). The relative risk of unexplained stillbirth among women with abnormal uterine artery Doppler was similar across the whole range of gestation. Overall, 4,769 (15.6%) of women had abnormal Doppler (either mean pulsatility index in the top decile or bilateral notches). Among these women, the absolute risks per 1,000 (95% confidence interval of stillbirth were 5.9 (3.9–8.5) up to 32 weeks and 3.1 (1.8–5.2) at or after 33 weeks of gestation. The risks of these events for the whole population were 1.3 (1.0–1.8) and 2.0 (1.5–2.6), respectively. The relative risk of stillbirth associated with abnormal Doppler did not significantly differ across the seven centers in relation either to all-cause stillbirth (P=.2) or stillbirth due to a placental cause (P=.3).
|
|
The ability of combinations of Doppler and maternal indices to predict stillbirth risk were then explored in relation to all-cause stillbirth, separately examining events before 33 weeks of gestation and those at 33 weeks and beyond (Table 3). Uterine artery Doppler was a better predictor of stillbirth before 33 weeks of gestation than maternal characteristics. Conversely, maternal characteristics performed better at later gestations. When the population in the top 5% of predicted risk were regarded as having screened positive, combined models of maternal and Doppler characteristics had a sensitivity of approximately 50% for early stillbirths but only 20% for late stillbirths. Logistic regression models for each event were generated using selection of significant variables and then converted to likelihood ratios (Tables 4 and 5). These provide a means to estimate stillbirth risk on the basis of uterine Doppler and maternal characteristics.
|
|
|
|
DISCUSSION |
|---|
|
TOP
ABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The approach of characterizing an arbitrary threshold of mean pulsatility index as being elevated results in loss of information. Moreover, some maternal characteristics were also significantly predictive of the risk of stillbirth. Prediction of the risk of stillbirth for an individual women would ideally take into account the precise level of the mean pulsatility index and combine this with her other characteristics to produce an estimate of her personal risk. We addressed these issues by developing multivariable logistic regression models that incorporated mean pulsatility index as a continuous variable along with the maternal characteristics. Although this addresses the two issues above, estimating a summary probability from such a model requires very sophisticated statistical knowledge. We addressed this by using a recently described method to convert logistic regression models into tables of likelihood ratios.21 A summary likelihood ratio can be calculated by multiplying the combination of likelihood ratios which are associated with a given woman’s characteristics. The background risk of stillbirth is then multiplied by the summary likelihood ratio and the woman’s individual risk can be calculated. Before widespread application of these models, their accuracy should be evaluated in other populations. The fact that the association between abnormal uterine artery Doppler and the risk of stillbirth did not significantly vary across the seven centers suggests that these models are likely to be generalizable. Moreover, an advantage of the likelihood ratio approach is that it is relatively simple to account for variation in the background risk of stillbirth when applied to other populations.
The data from the present study are essential for the design of any interventional studies that aim to reduce the risk of stillbirth in an unselected population screened using uterine artery Doppler. A series of randomized controlled trials of uterine and umbilical artery Doppler have been conducted. Meta-analysis of these trials demonstrates no reduction in perinatal mortality.22 However, these trials were conducted in the absence of detailed information about the screening properties of the test. The key properties which need to be known when designing a trial are the range of gestational age where the risk is increased and the absolute risk of stillbirth over that period. The timing of the risk is essential for planning the timing of any intervention. The absolute risk of a loss associated with an abnormal test result is essential for a power calculation. Previous trials were conducted in the absence of these data.22 The current model could be used in a trial to identify women at increased risk of stillbirth. There are a number of possible interventions in the high-risk group. One example would be intensive fetal surveillance (computerized cardiotocography, biometry, uteroplacental and venous Doppler, and biophysical profile) with delivery of those deemed to be at imminent risk of intrauterine fetal death.
A relative weakness of the present analysis was that the Doppler results were revealed and clinical management was modified on the basis of the result. The effect of this on the present analysis is difficult to assess. The nature of the intervention was that women with an elevated uterine mean pulsatility index were scanned every 4 weeks. This is comparable to the protocol used in a randomized controlled trial of uterine artery Doppler that did not demonstrate a protective effect on perinatal mortality.23 However, it is possible that some fetal deaths may have been prevented by this intervention or by the attending obstetrician being aware of the abnormal uterine artery Doppler. Further studies with a noninterventional design would be required to eliminate this potential source of error.
A further weakness in the present study relates to classification of stillbirths on the basis of birth weight. If the interval between death of the fetus and birth is prolonged, maceration of the fetus may occur. Consequently, the birth weight of the stillborn infant may not accurately reflect growth in fetal life. Moreover, dichotomizing infants as small or appropriate for gestational age on the basis of birth weight does not identify those infants that were constitutionally small. Moreover, infants with an apparently appropriate birth weight for gestational age may not have achieved their true growth potential. Therefore, classification of unexplained stillbirths by birth weight does not perfectly discriminate between those that were growth restricted and those that were not. Indeed, failure of correct classification may account for the association between unexplained stillbirth and abnormal uterine artery Doppler. Further studies may be able to address this using more sophisticated indicators of intrauterine growth restriction, such as serial antenatal scanning or neonatal anthropometry.
In conclusion, we show that a high resistance pattern of flow in the uterine artery is associated with an increased risk of stillbirth. The association is strongest for stillbirths due to placental dysfunction and, since these tend to occur at earlier gestations, is strongest for stillbirths occurring at extreme preterm gestations. Uterine artery Doppler is a relatively poor predictor of unexplained stillbirth unrelated to fetal growth restriction.
|
Footnotes |
|---|
Corresponding author: Gordon C. S. Smith, MD PhD, Professor of Obstetrics and Gynaecology, Cambridge University, Box 223, Rosie Maternity Hospital, Cambridge, CB2 2SW, United Kingdom; e-mail: gcss2{at}cam.ac.uk.
doi:10.1097/01.AOG.0000248536.94919.e3
|
REFERENCES |
|---|
|
TOP
ABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2. Hey EN, Lloyd DJ, Wigglesworth JS. Classifying perinatal death: fetal and neonatal factors. Br J Obstet Gynaecol 1986;93:1213–23.[Medline]
3. M Kady S, Gardosi J. Perinatal mortality and fetal growth restriction. Best Pract Res Clin Obstet Gynaecol 2004;18: 397–410.[Medline]
4. Smith GC, Crossley JA, Aitken DA, Pell JP, Cameron AD, Connor JM, et al. First-trimester placentation and the risk of antepartum stillbirth. JAMA 2004;292:2249–54.
5. Lyall F. The human placental bed revisited. Placenta 2002;23:555–62.[Medline]
6. Lees C, Parra M, Missfelder-Lobos H, Morgans A, Fletcher O, Nicolaides KH. Individualized risk assessment for adverse pregnancy outcome by uterine artery Doppler at 23 weeks. Obstet Gynecol 2001;98:369–73.
7. Yu CK, Smith GC, Papageorghiou AT, Cacho AM, Nicolaides KH, Fetal Medicine Foundation Second Trimester Screening Group. An integrated model for the prediction of preeclampsia using maternal factors and uterine artery Doppler velocimetry in unselected low-risk women. Am J Obstet Gynecol 2005;193:429–36.[Medline]
8. Yu CK, Papageorghiou AT, Parra M, Palma Dias R, Nicolaides KH, Fetal Medicine Foundation Second Trimester Screening Group. Randomized controlled trial using low-doseaspirin in the prevention of pre-eclampsia in women with abnormal uterine artery Doppler at 23 weeks’ gestation. Ultrasound Obstet Gynecol 2003;22:233–9.[Medline]
9. To MS, Alfirevic Z, Heath VC, Cicero S, Cacho AM, Williamson PR, et al. Cervical cerclage for prevention of preterm delivery in women with short cervix: randomised controlled trial. Lancet 2004;363:1849–53.[Medline]
10. Albaiges G, Missfelder-Lobos H, Lees C, Parra M, Nicolaides KH. One-stage screening for pregnancy complications by color Doppler assessment of the uterine arteries at 23 weeks’ gestation. Obstet Gynecol 2000;96:559–64.
11. Yudkin PL, Wood L, Redman CW. Risk of unexplained stillbirth at different gestational ages. Lancet 1987;1:1192–4.[Medline]
12. Smith GC. Life-table analysis of the risk of perinatal death at term and post term in singleton pregnancies. Am J Obstet Gynecol 2001;184:489–96.[Medline]
13. Maller RA, Zhou X. Survival analysis with long-term survivors. New York (NY): John Wiley & Sons; 1996.
14. Pocock SJ, Clayton TC, Altman DG. Survival plots of time-to-event outcomes in clinical trials: good practice and pitfalls. Lancet 2002;359:1686–9.[Medline]
15. Hosmer DW Jr, Lemeshow S. Applied survival analysis: regression modeling of time to event data. New York (NY): John Wiley & Sons; 1999.
16. Grambsch PM, Therneau TM. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 1994;81:515–26.
17. Smith GC, Pell JP, Dobbie R. Caesarean section and risk of unexplained stillbirth in subsequent pregnancy. Lancet 2003;362:1779–84.[Medline]
18. Platt RW, Joseph KS, Ananth CV, Grondines J, Abrahamowicz M, Kramer MS. A proportional hazards model with time-dependent covariates and time-varying effects for analysis of fetal and infant death. Am J Epidemiol 2004;160:199–206.
19. Royston P, Altman DG. Regression using fractional polynomials of continuous covariates: parsimonious parametric modelling. Appl Stat 1994;43:429–67.
20. Hosmer DW, Lemeshow S. Applied logistic regression. 2nd ed. New York (NY): John Wiley & Sons; 2000.
21. Smith GC, White IR, Pell JP, Dobbie R. Predicting cesarean section and uterine rupture among women attempting vaginal birth after prior cesarean section. PloS Med 2005;2:e252.[Medline]
22. Bricker L, Neilson JP. Routine Doppler ultrasound in pregnancy. Cochrane Database Syst Rev 2000;2:CD001450.[Medline]
23. Davies JA, Gallivan S, Spencer JA. Randomised controlled trial of Doppler ultrasound screening of placental perfusion during pregnancy. Lancet 1992;340:1299–303.[Medline]
This article has been cited by other articles:
|
|
 |
|
  U. M. Reddy Prediction and Prevention of Recurrent Stillbirth Obstet. Gynecol., November 1, 2007; 110(5): 1151 - 1164. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
