Evidence of early first-trimester growth restriction in pregnancies that subsequently end in miscarriage


  • F Mukri,

    Corresponding author
    1. Early Pregnancy, Gynaecological Ultrasound and MAS Unit, St George’s Hospital, London, UK
      Dr F Mukri, Early Pregnancy, Gynaecological Ultrasound and MAS Unit, St George’s Healthcare NHS Trust, Blackshaw Road, London SW17 0QT, UK. Email faizah_mukri@hotmail.com
    Search for more papers by this author
  • T Bourne,

    1. Early Pregnancy, Gynaecological Ultrasound and MAS Unit, St George’s Hospital, London, UK
    2. St George’s University of London, London, UK
    3. Department of Obstetrics and Gynaecology, University Hospital Gasthuisberg, KU Leuven, Belgium
    Search for more papers by this author
  • C Bottomley,

    1. Early Pregnancy, Gynaecological Ultrasound and MAS Unit, St George’s Hospital, London, UK
    2. St George’s University of London, London, UK
    Search for more papers by this author
  • C Schoeb,

    1. St George’s University of London, London, UK
    Search for more papers by this author
  • E Kirk,

    1. Early Pregnancy, Gynaecological Ultrasound and MAS Unit, St George’s Hospital, London, UK
    Search for more papers by this author
  • AT Papageorghiou

    1. Early Pregnancy, Gynaecological Ultrasound and MAS Unit, St George’s Hospital, London, UK
    2. St George’s University of London, London, UK
    3. Fetal Medicine Unit, St George’s Hospital, London, UK
    Search for more papers by this author

Dr F Mukri, Early Pregnancy, Gynaecological Ultrasound and MAS Unit, St George’s Healthcare NHS Trust, Blackshaw Road, London SW17 0QT, UK. Email faizah_mukri@hotmail.com


Objectives  To examine whether viable early pregnancies that subsequently end in miscarriage exhibit evidence of first-trimester growth restriction.

Design  Prospective cohort study.

Setting  Early pregnancy unit (EPU) of a teaching hospital.

Population  Women attending EPU between 5 and 10 weeks of gestation.

Methods  Women with spontaneously conceived intrauterine, viable singleton pregnancies with certain last menstrual period and regular cycles were included. The deviation between the observed and expected crown–rump length (CRL) for gestation was calculated and expressed as a z score. Pregnancies were followed up until the 11–14 week scan, and the deviation between those that remained viable and miscarried subsequently was calculated.

Main outcome measures  Viability at 11–14 week scan.

Results  Over 6 months, 316 women met the inclusion criteria. Twenty-four (7.4%) women were excluded. Of the remaining 292, the pregnancy remained viable in 251 (86%) and 41 (14%) suffered a miscarriage. At the first transvaginal ultrasound, the z score of the mean measured CRL for pregnancies that remained viable was −0.82, SD 1.46, while in pregnancies that subsequently miscarried the z score was −2.42 and the CRL was significantly smaller, SD 1.31 (P < 0.0001). In the latter group, the initial CRL was below the expected mean for gestational age in all women, while in 61% (25/41), the CRL was at least 2 SDs below the expected mean.

Conclusions  CRL was significantly smaller in pregnancies that subsequently ended in miscarriage. This suggests that early first-trimester growth restriction is associated with subsequent intrauterine death.


Transvaginal ultrasound (TVS) is a commonly performed investigation during early pregnancy and is used to confirm the pregnancy location, viability and gestation. Measurement of embryonic crown–rump length (CRL) is conventionally performed to confirm the gestational age of the pregnancy. The diagnosis of miscarriage is made in at least 10–20% of pregnancies in the first trimester.1 However, in pregnancies where fetal viability is demonstrated, the rate of subsequent miscarriage is lower (2–16%) but depends on the population studied, the indication for the scan and gestational age.2–4 Previous studies have evaluated various factors in the prediction of pregnancy outcome. Ultrasound findings that have been associated with subsequent miscarriage include a slow embryonic heart rate, a small gestational sac diameter and a larger-than-expected yolk sac diameter.5,6 Smaller-than-expected CRL has also been associated with subsequent pregnancy loss in studies involving women with threatened miscarriage and in women who have undergone assisted conception techniques.7–9 One possible link between a small CRL and subsequent miscarriage is that chromosomal defects not only have a high intrauterine lethality rate10 but can also be associated with fetal growth restriction in the first trimester of pregnancy.11,12 At what gestation this becomes apparent is not known yet.

The aim of this study was to prospectively assess whether spontaneously conceived, viable singleton pregnancies from an early pregnancy unit (EPU) population, which subsequently end in miscarriage, exhibit a different pattern of growth from those that remain viable.


This was a prospective cohort study carried out in the EPU of a large teaching hospital. This is a walk-in service where women are seen after self-referral or referral by their GP, midwife or the emergency department. All women attending the unit were offered the option of being included in this prospective observational study. Those who agreed were asked to complete a questionnaire including demographic features, detailed obstetric and menstrual histories and pregnancy symptoms. All women underwent a TVS using a 5-MHz transducer for B-mode imaging (Aloka SSD 900, 2000, 4000; Aloka, Tokyo, Japan) as part of the pregnancy assessment to determine pregnancy location, viability and gestation. In viable pregnancies, the number of embryos was documented and measurement of embryonic CRL performed. All women with viable singleton intrauterine pregnancies at 5–10 completed weeks of gestation, calculated from the last menstrual period (LMP), were included. To allow accurate assessment of gestation, the inclusion criteria used were as follows: certain dates (known LMP); regular menstrual cycles with a cycle between 26 and 30 days; and no hormonal contraception use, pregnancy or breastfeeding in the 3 months preceding the LMP. As embryonic growth may be different in pregnancies from assisted reproductive techniques,13,14 only those with spontaneous conception were included. Maternal characteristics (demographics, obstetric and menstrual history) and ultrasound findings were recorded in a computerised database (Viewpoint PIA database; LB Systems, Vienna, Austria) at the time of the assessment. Pregnancy outcome was then recorded when it became available. Outcomes were ascertained from subsequent scans recorded in the EPU or fetal medicine unit databases, or by contacting women or their GP. Outcome assessors were blinded to the CRL deviation at the time of the scan.

The primary outcome was continuing viability of the pregnancy beyond the first trimester, as determined by viability at the routine 11–14 week ultrasound assessment or known viability at a later gestation.

Statistical analysis

To compare the differences in CRL between those pregnancies that remained viable and those that subsequently miscarried, the deviation of observed and expected CRL was calculated. The observed CRL was established from the initial ultrasound examination. The expected CRL was calculated based on established growth curves of gestational age and CRL.15 These growth curves have been well validated using the transvaginal route.16,17 The difference between the observed and expected CRL was calculated and expressed in SDs (z score). Continuous variables between groups, including the z scores, were compared using a t test or Mann–Whitney U test for parametric and nonparametric data as appropriate, and a chi-square test for dichotomous variables. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for the prediction of subsequent pregnancy loss following a scan demonstrating a CRL below −2 SD from the expected mean were calculated. Two-sided P values are reported throughout.


Over the 6-month study period, 1486 women participated in the study. Of these, 316 (21.3%) met all the inclusion criteria. The median gestational age at ultrasound in these women was 7 weeks (interquartile range 6–8 weeks). Twenty-four women were excluded from the subsequent analysis: 8 (2.5%) women who opted for pregnancy termination after the first scan and 16 (5.1%) for whom first-trimester pregnancy outcome was not available. Women lost to follow up were more likely to be nulliparous when compared with those for whom outcome for pregnancy was available but no other statistically significant difference was found (Table 1).

Table 1.  Maternal characteristics between women with complete follow up and those lost to follow up
 Included in the study (n = 292)Lost to follow up (n = 16)P value
  1. IQR, interquartile range.

  2. Data are presented as n (%) unless indicated.

Age, median (IQR)33 (29–36)31 (29–35)0.35
Ethnicity 0.74
White186 (64)12 (75) 
Black46 (16)1 (6) 
Asian37 (13)2 (13) 
Other23 (8)1 (6) 
Parity 0.03
0155 (53)14 (88) 
192 (32)1 (6) 
>145 (15)1 (6) 
Previous miscarriage (weeks) 0.75
<12119 (41)5 (31) 
12–2415 (5)1 (6) 
Indication 0.46
Bleeding73 (25)2 (13) 
Pain65 (22)3 (19) 
Bleeding and pain70 (24)7 (44) 
Anxiety/previous history69 (24)3 (19) 
Other15 (5)1 (6) 

Of the remaining 292 women, the pregnancy remained viable at 11–14 weeks in 251 (86.0%) and 41 (14.0%) suffered a miscarriage between their initial presentation and the 11–14 week scan. There was no statistically significant difference between maternal characteristics, parity or obstetric history between those women who subsequently suffered a miscarriage when compared with women in whom pregnancies remained viable. In addition, no significant difference was seen in the likelihood of miscarriage between those initially presenting with pain alone, bleeding alone, pain and bleeding or neither pain nor bleeding (Table 2).

Table 2.  Maternal characteristics in women with and without subsequent miscarriage
 Remain viable (n = 251)Subsequent miscarriage (n = 41)P value
  1. IQR, interquartile range.

  2. Data are presented as n (%) unless indicated.

Age, median (IQR)33 (30–37)33 (29–35)0.30
Ethnicity 0.57
White163 (65)23 (56) 
Black38 (15)8 (20) 
Asian32 (13)5 (12) 
Other18 (7)5 (12) 
Parity 0.44
0130 (52)25 (61) 
180 (32)12 (29) 
>141 (16)4 (10) 
Previous miscarriage (weeks) 0.35
<1299 (39)20 (49) 
12–2412 (5)3 (7) 
Indication 0.25
Bleeding62 (25)11 (27) 
Pain58 (23)7 (17) 
Bleeding and pain55 (22)15 (37) 
Anxiety/previous history62 (25)7 (17) 
Other14 (6)1 (2) 

The mean z score for CRL in pregnancies that remained viable was −0.82 (SD 1.46), while in those that subsequently miscarried, the mean z score for CRL was significantly larger at −2.42 (SD 1.31, t = 6.60, P < 0.0001, Figure 1). All pregnancies that subsequently miscarried had a CRL below the expected CRL for gestation, while in 61% (25/41), the CRL was 2 SDs or more below that expected for gestational age. In pregnancies that remained viable, the CRL was less than that expected in 70% (175/251), while in 22% (55/251), the CRL was 2 SDs or more below the expected CRL. There was a relationship between increasing CRL z score and miscarriage, with the risk of miscarriage increasing with increasing CRL deviation (Table 3). Using a CRL measurement below 2 SDs as a cutoff in predicting subsequent miscarriage yielded a sensitivity of 61%, with a specificity of 78.1%, PPV of 31.3% and NPV of 92.5%.

Figure 1.

Expected and observed CRL in the study group. The solid line and dashed lines are the expected median CRL 2 SDs for gestational age, respectively. The blue triangles represent CRL of pregnancies subsequently viable at 11–14 weeks. The red triangles show CRL of pregnancies that were viable at initial early first-trimester TVS but subsequently miscarried.

Table 3.  Deviation of CRL on the first scan and subsequent outcome
z scoreRemain viable (n)Subsequent miscarriage (n)Percent subsequent miscarriage (%)
−2 or more196168
−3 to less than −2401122
−4 to less than −313838
Less than −42675

There were 16 women with a history of recurrent miscarriage, defined as three or more consecutive miscarriages. Of these, 11 had a pregnancy that remained viable at 11–14 weeks and 5 had a subsequent pregnancy failure. Analysis excluding these women showed that the differences between mean z score in pregnancies that remained viable (−0.78) compared with those that subsequently miscarried (−2.55) was still significant (t = 6.64, P < 0.0001).


In this study, we have demonstrated that in women with spontaneously conceived singleton pregnancies and certain menstrual dates who attend an EPU, the CRL is below that expected for gestation in all cases that will subsequently miscarry, and is at least 2 SD below that expected in almost two-thirds of women. Seventy percent of the continuing pregnancies also had a CRL, which was less than that expected by gestational age, despite women only having been included if they had certain dates and regular cycles. This phenomenon of women overestimating gestational age has been reported before and is likely to be attributed to late ovulation or inaccurate reporting of LMP date,18 although it may be that charts based on apparent LMP in fact underestimate gestational age.19 However in viable continuing pregnancies, the mean deviation between the observed and expected CRL was 22%, while it was 61% in those pregnancies that subsequently miscarried.

The demonstration of positive fetal heart action with early TVS is generally perceived as a reassuring finding for many women. However, despite the pregnancy being viable at the time of the initial TVS, a smaller-than-expected CRL may be an indicator of early fetal growth delay. It is known that at least 70% of first-trimester losses are due to chromosomal abnormalities20 and that these can be associated with early growth delay.10 However, a study by Bessho et al. showed that the growth delay in fetuses prior to miscarriage was similar in both fetuses with normal and abnormal karyotype.21 Thus, although the reason for the slower rate of fetal growth might be attributable to chromosomal abnormality in the majority, there may be another aetiology that affects karyotypically normal fetuses that miscarry. One possibility is that abnormal placentation could play a role; histological examination in cases of early pregnancy loss has shown a thinner and more fragmented trophoblastic shell and reduced cytotrophoblast invasion of spiral arteries than in normal pregnancy, which eventually leads to indirect oxidative stress effects and cellular dysfunction.22 It is possible that embryonic growth could be affected through this process, representing an early manifestation of placental dysfunction.

In our study, the rate of miscarriage after visualisation of a positive fetal heart action was 14%. This is comparable to some studies,3 although other studies have shown lower rates.2 These differences may be partly explained by the earlier gestational age at presentation in our study. Although the population in our study was unselected, about half had some vaginal bleeding at the initial presentation. While a study by Tongsong et al.23 demonstrated that the rate of miscarriage after visualisation of fetal heart is higher in women with threatened miscarriage when compared with a control group without bleeding (5.5 versus 1.9%), in our study, we did not find that women were more likely to have an early miscarriage if they presented with bleeding, once the fetal heart was visualised in early pregnancy.

It is common practice to amend the estimated date of delivery for the pregnancy if there is a discrepancy of more than 7 days between dates according to the LMP and the ultrasound findings, even in women who are certain of their last period. This is because accurate dating in early pregnancy decreases inappropriate ‘postterm’ induction of labour.24,25 Our study suggests that despite many women overestimating their gestational age, smaller-than-expected CRL may also be a sign of early fetal growth delay. We have shown that the risk of subsequent miscarriage increases with increasing deviation from the expected CRL. Thus, while the risk of miscarriage is 8% in those with CRL within 2 SDs, the risk is 22, 38 and 75% for those with a CRL between −2 and −3 SDs, −3 and −4 SDs and less than −4 SDs, respectively (Table 3). It may therefore be reasonable to offer women further interval ultrasound assessment to confirm continuing viability. Whether a policy of re-dating or interval ultrasound is adopted will depend on resource allocation, and units may adopt a policy of arranging follow up for those in whom the initial ultrasound shows the largest deviation. Larger longitudinal studies are needed to obtain more detailed clinical recommendations regarding counselling of such women and when the pregnancy should be reassessed rather than simply re-dated. Although performing such a repeat examination in this group of women is unlikely to affect the ultimate outcome of the pregnancy, recognition that smaller-than-expected CRL is a marker of early pregnancy failure may aid in counselling and reduce the number of women who are found to have unrecognised failed pregnancies at the time of the 11–14 week ultrasound assessment, with its attendant psychological morbidity.26,27

One of the limitations of the study is that, although women were strictly selected by certainty of their LMP, there is still a small possibility that the gestational age was not accurately estimated in all women. Pregnancies with smaller CRL may therefore be of earlier gestation, which is associated with a higher miscarriage rate. Nevertheless, the strict inclusion criteria we used relating to the certainty of the LMP mean that deviation of CRL is may be more likely to be due to growth delay. One way of overcoming this limitation is the use of pregnancies resulting from in vitro fertilisation (IVF) this has been used in the past, but as embryonic growth rate in such pregnancies may be different from that in normal conception, we excluded women who had undergone assisted conception in the current pregnancy.


This prospective study shows that early first-trimester growth restriction increases the risk of spontaneous fetal demise. In women with certain LMP, CRL for gestation was smaller in pregnancies that subsequently ended in miscarriage than in those that remained viable. All pregnancies that ended in miscarriage had a CRL below that expected for gestation, and in almost two-thirds, the CRL was at least 2 SD below expected CRL. It is important to consider whether pregnancy abnormality and imminent fetal demise account for a significant discrepancy between observed and expected CRL in this population of women. Simple re-dating of such pregnancies may provide false reassurance to women.

Contribution to authorship

F.M. wrote the original draft of the manuscript. T.B. and A.T.P. designed and supervised the study. C.B. and F.M. collected the data. A.T.P. analysed and interpreted the data. C.S. and E.K. helped with design and data collection. C.B., A.T.P., C.S., E.K., and T.B. revised the manuscript.

Details of ethics approval

Confirmed by Local Research Ethics Committee that ethics approval is not required for this observational study.


The authors thank early pregnancy unit staff at St George’s Hospital for their help in recruiting women for the study.

Journal club

Background What are the first trimester sonographic risk factors associated with subsequent miscarriage? By what mechanisms would a small crown rump length (CRL) be associated with subsequent miscarriages?

Technical Do you think this study has recruited a patient population that would be comparable to the one seen in your unit? How reliable do you think last menstrual period (LMR) dating is?

Clinical practice Based on the finding of this study, how will you assess dating in the first trimester when the ‘CRL’ gestational age is discrepant from the ‘LMP’ gestational age? How would you go about developing a guideline for dating by ultrasound and LMP in the first trimester? What would be your criteria for correcting LMP dates? How would you counsel and follow-up women with a significant LMP/ultrasound (US) dates discrepancy?

Future research How would you design a prospective study to confirm the findings of this report? What patient population would you want to study, and how large a sample size?

Correspondence: Dr V Berghella, Department of Obstetrics and Gynecology, Division of Maternal-Fetal Medicine, Thomas Jefferson University, 834 Chestnut Street, Suite 400, Philadelphia, PA 19107, USA.
Email vincenzo.berghella@jefferson.edu