Human Reproduction, Vol.29, No.5 pp.

931 937, 2014

Advanced Access publication on March 6, 2014 doi:10.1093/humrep/deu042

ORIGINAL ARTICLE Early pregnancy

Non-visualized pregnancy losses

are prognostically important for
unexplained recurrent miscarriage
A.M. Kolte 1,*, R.H. van Oppenraaij 2, S. Quenby 3, R.G. Farquharson 4,
M. Stephenson5, M. Goddijn 6, and O.B. Christiansen1,7 on behalf of the
ESHRE Special Interest Group Early Pregnancy
1
Recurrent Miscarriage Unit, Fertility Clinic 4071, University Hospital Copenhagen, Rigshospitalet, Blegdamsvej 9, DK-2100 Kbenhavn ,
Denmark 2Department of Obstetrics and Gynaecology - Subdivision Obstetrics & Prenatal Care, Erasmus MC, Rotterdam, The Netherlands
3
Clinical Sciences Research Institute, University Hospital Coventry, Warwick Medical School, Warwick, UK 4Department of Obstetrics and
Gynaecology, Liverpool Womens Hospital, Liverpool, UK 5Department of Obstetrics and Gynecology, University of Illinois at Chicago, Chicago,
IL, USA 6Center for Reproductive Medicine, Department of Obstetrics and Gynaecology, Academic Medical Center, Amsterdam,
The Netherlands 7Department of Obstetrics and Gynaecology, Aalborg Hospital, Aalborg, Denmark

*Correspondence address. Recurrent Miscarriage Unit, Fertility Clinic 4071, University Hospital Copenhagen Rigshospitalet, Blegdamsvej 9,
DK-2100 Kbenhavn , Denmark. Tel: +45-3545-4951; E-mail: astrid.marie.kolte@regionh.dk

Submitted on December 20, 2013; resubmitted on February 3, 2014; accepted on February 5, 2014

study question: Are non-visualized pregnancy losses (biochemical pregnancy loss and failed pregnancy of unknown location combined) in the
reproductive history of women with unexplained recurrent miscarriage (RM) negatively associated with the chance of live birth in a subsequent pregnancy?
summaryanswer: Non-visualized pregnancy losses contribute negatively to the chance for live birth: each non-visualized pregnancy loss confers
a relative risk (RR) for live birth of 0.90 (95% CI 0.83; 0.97), equivalent to the RR conferred by each additional clinical miscarriage.
what is known already: The number of clinical miscarriages prior to referral is an important determinant for live birth in women with RM,
whereas the signicance of non-visualized pregnancy losses is unknown.
study design, size, duration: A retrospective cohort study comprising 587 women with RM seen in a tertiary RM unit 20002010. Data
on the outcome of the rst pregnancy after referral were analysed for 499 women.
participants/materials, setting, methods: The study was conducted in the RM Unit at Rigshospitalet, Copenhagen, Denmark.
We included all women with unexplained RM, dened as 3 consecutive clinical miscarriages or non-visualized pregnancy losses following spontaneous
conception or homologous insemination. The category non-visualized pregnancy losses combines biochemical pregnancy loss (positive hCG, no ultra-
sound performed) and failed PUL (pregnancy of unknown location, positive hCG, but on ultrasound, no pregnancy location established). Demographics
were collected, including BMI, age at rst pregnancy after referral and outcome of pregnancies prior to referral. Using our own records and records from
other Danish hospitals, we veried the outcome of the rst pregnancy after referral. For each non-visualized pregnancy loss and miscarriage in the
womens reproductive history, the RR for live birth in the rst pregnancy after referral was determined by robust Poisson regression analysis, adjusting
for risk factors for negative pregnancy outcome.
main results and the role of chance: Non-visualized pregnancy losses constituted 37% of reported pregnancies prior to referral
among women with RM. Each additional non-visualized pregnancy loss conferred an RR for live birth of 0.90 (95% CI 0.83; 0.97), which was not statistically
signicantly different from the corresponding RR of 0.87 (95% CI 0.80; 0.94) conferred by each clinical miscarriage. Among women with 2 clinical
miscarriages, a reduced RR for live birth was also shown: 0.82 (95% CI 0.74; 0.92) for each clinical miscarriage and 0.89 (95% CI 0.80; 0.98) for each
non-visualized pregnancy loss, respectively. Surgically treated ectopic pregnancies (EPs) were signicantly more common for women with primary
RM and no conrmed clinical miscarriages, compared with women with primary RM and 1 clinical miscarriage (22 versus 6%, difference 16% (95%
CI 9.1%; 28.7%); RR for ectopic pregnancy was 4.0 (95% CI 1.92; 8.20).
limitations, reasons for caution: RM was dened as 3 consecutive pregnancy losses before 12 weeks gestation, and we included
only women with unexplained RM after thorough evaluation. It is uncertain whether the ndings apply to other denitions of RM and among women with
known causes for their miscarriages.

& The Author 2014. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved.
For Permissions, please email: journals.permissions@oup.com

Downloaded from https://academic.oup.com/humrep/article-abstract/29/5/931/675614/Non-visualized-pregnancy-losses-are-prognostically

by E-nformation user
on 03 October 2017
 932 Kolte et al.

wider implications of the findings: To our knowledge, this is the rst comprehensive investigation of prior non-visualized pregnancy
losses and their prognostic signicance for live birth in a subsequent pregnancy in women with unexplained RM. We show that a prior non-visualized
pregnancy loss has a negative prognostic impact on subsequent live birth and is thus clinically signicant.
study funding/competing interest(s): None.
trial registration number: N/A.
Key words: recurrent miscarriage / biochemical pregnancy loss / pregnancy of unknown location / non-visualized pregnancy loss /
retrospective cohort study

To investigate whether prior non-visualized pregnancy losses are clin-

Introduction ically relevant, we collected data over 10 years on the outcome of the rst
The term early pregnancy loss covers three different clinical scenarios: pregnancy after referral to the Danish Recurrent Miscarriage Unit.
miscarriage, where transvaginal ultrasound (TVS) or histological nd-
ings document an intrauterine demise before 12 weeks gestation;
ectopic pregnancy (EP), where TVS or laparoscopy identies a pregnancy Materials and Methods
outside the uterine cavity; and biochemical pregnancy loss, where there
A retrospective cohort study comprising 918 consecutive women seen in the
is a positive pregnancy test but no ultrasound has been performed. In the Danish RM Unit at the Fertility Clinic, University Hospital Copenhagen, Rig-
event where a woman has a positive pregnancy test and TVS is per- shospitalet from January 2000 to January 2011 was performed. We included
formed, but neither an intrauterine nor an ectopic pregnancy is only women who we considered as having unexplained RM, i.e. who fullled
seen, the pregnancy is classied as a pregnancy of unknown location the following criteria: at least three consecutive pregnancy losses, including
(PUL) (Barnhart et al., 2011). Following an initial classication of PUL, both clinical miscarriages and non-visualized pregnancy losses; age ,40
the possible nal diagnoses are: an ongoing intrauterine pregnancy; years at referral; regular menstrual cycle with length 23 35 days (variation
an ectopic pregnancy; a failed PUL; or an intrauterine miscarriage from cycle to cycle was 2 3 days); normal uterine evaluation by hystero-
(Barnhart et al., 2011). scopy, hysterosalpingogram or uterine hydrosonography; normal parental
When dealing with an acute early pregnancy complication, the distinc- karyotypes; and negativity for the lupus anticoagulant. We excluded
women who had conceived after IVF/ICSI or donor insemination prior to
tion between different types of early pregnancy loss is very important as it
referral. Figure 1 gives an overview of study ow. In short, 331 women were
has implications for the prognosis, treatment and follow-up of patients
excluded. Forty women (7%) were lost to follow-up and according to their
(Kirk et al., 2009; Barnhart et al., 2011). In contrast, when considering records, 48 women (9%) did not conceive after referral. Outcome of rst preg-
the past reproductive history of a patient referred to a recurrent miscar- nancy after referral was registered for 499 women: 290 with primary RM
riage (RM) unit, the importance of early pregnancy losses, such as failed (PRM) (58%) and 209 with secondary RM (SRM) (42%). Of these, 368 (74%)
PULs and/or biochemical pregnancy losses, has not been well studied. had experienced 2 clinical miscarriages and thus fullled the ASRM criteria
We hypothesize that biochemical pregnancy losses and failed PULs for recurrent pregnancy loss as well as the ESHRE/RCOG criteria.
share similar prognostic importance. Therefore, in addition to separate As is standard practice in this RM unit, at rst consultation, all women had
analyses for biochemical pregnancy losses and failed PULs, we group given a detailed written account of their reproductive history along with
these two diagnoses together as non-visualized pregnancy losses documentation on where, when and how their previous pregnancies had
dened as a pregnancy loss initially conrmed by a positive hCG, but been managed. Treatment regimens in a subsequent pregnancy varied
according to medical history, and included tender loving care (TLC) with
not visualized by TVS, if performed.
or without intravenous immunoglobulin (IVIg). Twenty-seven women
The denition of RM is controversial. A guideline from the European
received IVIg or placebo from 2008 to 2013 (NCT00722475). One
Society of Human Reproduction and Embryology (ESHRE), as well as hundred and ten women received IVIg in a non-randomized fashion,
the Royal College of Obstetricians and Gynaecologists (RCOG), before 2008 or being ineligible for the trial. These women had had at least
dene RM as three or more consecutive pregnancy losses (Jauniaux four early pregnancy losses or at least one unexplained late miscarriage
et al., 2006; Regan et al., 2011). However, the American Society for Re- and two early pregnancy losses.
productive Medicine (ASRM) Practice Committee denes recurrent Patients were followed at the RM Unit until 16 weeks gestation, after
pregnancy loss as two or more clinical miscarriages conrmed by ultra- which the women were referred for continued monitoring at their local hos-
sound or histology, not necessarily consecutive (ASRM Practice Com- pital. Information on outcome on rst pregnancy after referral was obtained
mittee, 2013). Non-visualized pregnancy losses are thus not included either from patient records or from the women themselves.
in the ASRM Practice Committee denition, nor in other recent publica- For this study, the womens information was entered in a Microsoft Ofce
Access 2010 database by two of the authors (A.M.K. and O.B.C.). Double
tions (Saravelos and Li, 2012).
entry was avoided using the unique Danish identication number. Prior to
Non-visualized pregnancy losses in women with RM are increasingly
statistical analysis, data quality was checked manually (A.M.K.).
diagnosed because very early pregnancy testing is readily available We divided early pregnancy events into the following categories: miscar-
(Wilcox et al., 1987). Whether non-visualized pregnancy losses should riage, where ultrasound or histology documented an intrauterine pregnancy
be included in the denition criteria for RM is controversial. If they nega- loss before 12 weeks gestation; EP, where a pregnancy loss was visualized
tively affect the chance of a subsequent live birth, then non-visualized outside the uterus by laparoscopy or TVS; failed PUL, where there had
pregnancy losses are clinically relevant. been a positive hCG, but no location was established by TVS; biochemical

Downloaded from https://academic.oup.com/humrep/article-abstract/29/5/931/675614/Non-visualized-pregnancy-losses-are-prognostically

by E-nformation user
on 03 October 2017
 Early pregnancy losses and recurrent miscarriage 933

Figure 1 Inclusion of women in the cohort. All women were seen between January 2000 and January 2011. PRM, primary recurrent miscarriage; SRM,
secondary recurrent miscarriage; NVPL, non-visualized pregnancy loss.

pregnancy loss, as a positive hCG, but no TVS performed. The two categories
failed PUL and biochemical pregnancy loss were combined as non-
Results
visualized pregnancy losses. As the study is retrospective, all diagnoses are
nal.
Reproductive history
For 88% of the women in the cohort, we relied on self-reporting and Of 2781 pregnancies reported at rst consultation, 327 were births
records available at time of referral. However, as a data quality check, we after Week 22 (12%). Of the 2454 pregnancy losses, there were 1426
obtained further details on reported non-visualized pregnancy losses for miscarriages before Week 12 (58%), 578 biochemical pregnancy
the 61 women born on the rst to third of each month. losses (23%) and 334 (16%) failed PULs. Thus non-visualized pregnancy
losses constituted 37% of all pregnancy losses before referral in this group.
Additionally, there were 73 late miscarriages between Week 12 and 22
(3%) and 43 EPs (2%) (see Fig. 2). All EPs had been treated surgically.
Statistics Figure 3 shows the distribution of biochemical pregnancy losses, failed
As we have chosen to report our results as relative risk (RR) and the outcome PULs and miscarriages by gestational age. The mean gestational age for
live birth was common (.10%), we used robust Poisson regression instead
biochemical pregnancy losses was 6.08 weeks (95% CI for the mean
of standard logistic regression (Deddens and Petersen, 2008). In the Poisson
5.96; 6.19) and for failed PULs 6.59 (95% CI 6.43; 6.75). The difference
regression analysis we used non-visualized pregnancy losses as the independ-
ent variable and corrected for the risk factors PRM versus SRM; age at index is 0.51 weeks (95% CI 0.33; 0.70). The mean gestational age for clinical
pregnancy; the number of prior early and late miscarriages; EPs; and treat- miscarriages was 8.87 (95% CI 8.74; 9.01), signicantly higher than for
ment. Equivalent analyses were performed with non-visualized pregnancy non-visualized pregnancy losses, mean difference 2.60 weeks (95% CI
losses split into biochemical pregnancy losses and failed PULs. We also 2.44; 2.76).
used miscarriage as independent variable equivalent to non-visualized preg- As shown in Fig. 4, women with PRM and no clinical miscarriages had a
nancy loss. As standard Poisson regression uses the log-link function, female statistically signicantly higher frequency of surgically treated EPs than
age in years, early miscarriages, non-visualized pregnancy losses, biochemical those with at least one clinical miscarriage (22 versus 6%, difference
pregnancy losses, failed PULs and EPs were modelled as linear variables on 16% (95% CI 9.1%; 28.7%), corresponding to an RR for having had an
the logit scale. Testing for linearity showed no problems for any of the vari- EP of 4.0 (95% CI: 1.92; 8.20) in the former group. We did not
ables. Model control was performed. There were no signs of interaction
conrm the nding for women with SRM.
for any of the variables and thus multiple regression analysis was deemed ap-
The women for whom we attempted to obtain further details about
propriate. For these analyses the statistical software package STATA 11 was
used.
prior non-visualized pregnancy losses reported a total of 123 non-
Fishers exact test was used to test the hypothesis of equal proportions of visualized pregnancy losses, of which 77 (63%) were biochemical preg-
ectopic pregnancies (EPs) between different groups of patients. T-test was nancy losses and 46 (37%) were failed PULs. We were able to conrm
used for comparison of gestational age between groups of pregnancy loss. the self-reported information in all cases except one; the woman
For these analyses we used the statistical software package SAS 19.2. reported a biochemical pregnancy loss, which actually was a miscarriage.

Downloaded from https://academic.oup.com/humrep/article-abstract/29/5/931/675614/Non-visualized-pregnancy-losses-are-prognostically

by E-nformation user
on 03 October 2017
 934 Kolte et al.

Figure 3 Percentages of biochemical pregnancy losses, failed PULs

and miscarriages according to gestational age. PULs, pregnancies of
Figure 2 Number and type of pregnancy losses (reproductive unknown location.
history) reported by 587 RM women at rst consultation. In addition,
269 births after 22 weeks gestation were reported. EP: surgically
treated ectopic pregnancy; late miscarriage: intrauterine pregnancy
loss after 12 weeks gestation; early miscarriage: histologically or ultra-
sonically conrmed intrauterine pregnancy loss before 12 weeks gesta-
tion; biochemical pregnancy loss: positive hCG, no ultrasound
performed; failed PUL: failed pregnancy of unknown location, positive
hCG, but location not established by ultrasound.

Thus in 99% of cases, the self-reported information of non-visualized

pregnancy losses was conrmed.

Importance of reproductive history on live

birth
When analysing all 499 women in the cohort, the RR for live birth for each
non-visualized pregnancy loss was 0.90 (95% CI 0.83; 0.97) and for each
clinical miscarriage 0.87 (95% CI 0.8; 0.94). For biochemical pregnancy
losses the RR for live birth was 0.89 (95% CI 0.82; 0.97) and for failed
PULs 0.91 (95% CI 0.82; 1.02). We found no statistically signicant differ- Figure 4 Frequency of a history of surgically treated ectopic pregnan-
ence between the RRs for live birth conferred by each non-visualized cies (EPs) according to presence or absence of conrmed miscarriages
pregnancy loss and each miscarriage in any of the analyses. in the history, among 587 women; 331 with primary and 256 with sec-
For women with 2 clinical miscarriages, the RR for live birth was 0.89 ondary recurrent miscarriage. EP: surgically treated ectopic pregnancy:
(95% CI 0.80; 0.98) for non-visualized pregnancy loss and 0.82 (95% CI miscarriage, histologically or ultrasonically conrmed intrauterine preg-
0.74; 0.92) for clinical miscarriage and for biochemical pregnancy loss and nancy loss before 12 weeks gestation; NVPLs: non-visualized pregnancy
losses; biochemical pregnancy losses and failed pregnancies of unknown
failed PUL, the RR was 0.88 (95% CI 0.79; 0.98) and 0.89 (0.77; 1.04),
location combined.
respectively.
From Table I we noted that increasing age at rst pregnancy after re-
ferral was a small, but consistently signicant negative prognostic factor in
almost all subgroups with RR for live birth ranging from 0.97 to 0.99 for change in RR for live birth and BMI in itself had no signicant effect on
each additional year. outcome, neither as a continuous variable nor as a grouped variable
When limiting the Poisson regression analysis to the 312 women (BMI,20; 20 24; 25 29; 30; with BMI 2024 as reference) (see
(63%) for whom we had registered BMI, there was no signicant Table I).

Downloaded from https://academic.oup.com/humrep/article-abstract/29/5/931/675614/Non-visualized-pregnancy-losses-are-prognostically

by E-nformation user
on 03 October 2017
 Early pregnancy losses and recurrent miscarriage 935

Table I Relative risk (95% CI) of live birth in the index pregnancy, unadjusted for treatment.

Variable All women Women with 2 Women receiving

clinical miscarriages tender loving care
.............................................................................................................................................................................................
Pregnancy losses N 499 N 368 N 344
Age at rst pregnancy after referral 0.98 (0.96; 0.99) 0.98 (0.96; 1.00) 0.97 (0.96; 0.99)
Prior miscarriage 0.87 (0.80; 0.94) 0.82 (0.74; 0.92) 0.86 (0.78; 0.96)
Prior non-visualized pregnancy lossa 0.90 (0.83; 0.97) 0.89 (0.80; 0.98) 0.90 (0.82; 1.00)
Prior biochemical pregnancy loss 0.89 (0.82; 0.97) 0.88 (0.79; 0.98) 0.92 (0.83; 1.02)
Prior failed pregnancy of unknown location 0.91 (0.82; 1.02) 0.89 (0.77; 0.04) 0.84 (0.71; 0.99)
BMI N 312 N 228 N 221
BMI 20 1.13 (0.90; 1.41) 1.28 (1.00; 1.63) 1.00 (0.77; 1.31)
BMI 2125 1.00 (reference) 1.00 (reference) 1.00 (reference)
BMI 2629 1.02 (0.77; 1.35) 1.00 (0.70; 1.43) 0.85 (0.58; 1.24)
BMI 30 1.04 (0.81; 1.35) 1.06 (0.79; 1.43) 0.97 (0.70; 1.33)
a
Non-visualized pregnancy loss: combines biochemical pregnancy losses and failed pregnancies of unknown location.

We analysed the 290 PRM and 209 SRM women separately and the RR the majority of failed PULs are early intrauterine miscarriages (Kirk
did not change signicantly, but as expected, the condence intervals et al., 2009).
widened due to smaller numbers in each subgroup. Data on PRM When interpreting our results it is important to note that the denition
versus SRM women are available online as Supplementary data, Table SI. of RM applied in the study was three or more consecutive early preg-
nancy losses. Even though our calculations were based on a linear
model on the logit scale, we are unable to project the results to sporadic
Treatment pregnancy losses or other denitions of RM, e.g. two consecutive or non-
In an analysis of RR for live birth by non-visualized pregnancy loss, clinical consecutive early pregnancy losses. We did show a statistically signicant
miscarriage and biochemical pregnancy loss and failed PULs among the reduction in relative risk for live birth also for women who full the ASRM
344 women who received only TLC we found comparable results as denition of recurrent pregnancy loss, i.e. 2 clinical miscarriages. The
for the total group of 499 patients, as can be seen in Table I. ndings in this study should prompt further inquiry into an evidence-
Ninety-eight women received IVIg in addition to TLC and had an RR based denition of RM.
for live birth of 1.27 (95% CI 1.07; 1.52) compared with TLC alone, As the cohort only included women with unexplained RM, our results
and the RR of IVIg for live birth for patients with PRM was 1.39 (1.10; may not apply to other groups of patients, such as patients with chromo-
1.76). Adjustment for treatment did not signicantly alter the effect of somal abnormalities, irregular menstrual cycles and patients with RM
non-visualized pregnancy losses, miscarriages, biochemical pregnancy after IVF/ICSI.
losses and failed PULs on the RR for live birth (see Table II). To our knowledge, there are no data regarding the cost-effectiveness
of RM evaluation or treatment if non-visualized pregnancy losses are
included in the denition, although this is already clinical practice in
Discussion several European countries such as Great Britain and Denmark. As
resources in clinical care are limited, this would be a logical next step
Non-visualized pregnancy losses represent a signicant proportion of the
by health care economists and relevant policy makers.
pregnancy losses experienced by women referred to the Danish RM
clinic. We have demonstrated that non-visualized pregnancy losses
and miscarriages both have a negative prognostic inuence on the
chance for live birth in the rst pregnancy after referral among women Gestational age
with unexplained RM. The number of clinical miscarriages before referral The exclusion of non-visualized pregnancy losses in RM denitions is
has been reported to be an important determinant for RM womens probably based on reports that a transiently positive pregnancy test at
prognosis for live birth (Brigham et al., 1999; Lund et al., 2012). To the time of menstrual period is a common nding in normal women
our knowledge, this is the rst investigation of the prevalence and prog- (Wilcox et al., 1987). It is therefore noteworthy that the mean gestational
nostic signicance of non-visualized pregnancy losses in women with age at time of diagnosis of both biochemical pregnancy losses and failed
RM. PULs in this study was 6 weeks.
Presently, non-visualized pregnancy losses in the history of women The gestational age for non-visualized pregnancy losses is based on last
with RM are largely ignored by gynaecologists and general practitioners. menstrual period and may therefore be unreliable. However, as all
The nding that non-visualized pregnancy losses and early miscarriages women in the cohort had regular menstrual cycles with a variation of
have a similar negative effect on RR for live birth is thus very important. no more than 23 days for each individual woman, we assume that
At least for women with RM, our ndings support the assumption that the estimate of gestational age in prior pregnancies is reasonably

Downloaded from https://academic.oup.com/humrep/article-abstract/29/5/931/675614/Non-visualized-pregnancy-losses-are-prognostically

by E-nformation user
on 03 October 2017
 936 Kolte et al.

Table II Relative risk (95% CI) of live birth in the index pregnancy, adjusted for treatment.

Variable N All women N Women with 2 clinical

miscarriages
.............................................................................................................................................................................................
All women
Age at rst pregnancy after referral 499 0.98 (0.96; 0.99) 368 0.98 (0.96; 1.00)
Prior miscarriage 0.86 (0.79; 0.93) 0.81 (0.72; 0.91)
Prior non-visualized pregnancy lossa 0.89 (0.82; 0.96) 0.88 (0.79; 0.97)
Prior biochemical pregnancy loss 0.88 (0.81; 0.97) 0.88 (0.79; 0.98)
Prior failed pregnancy of unknown location 0.90 (0.80; 1.00) 0.87 (0.74; 1.02)
Tender loving care 344 1 (reference) 243 1.00 (reference)
Tender loving care and IvIg alone 98 1.28 (1.07; 1.52) 77 1.36 (1.11; 1.67)
Tender loving care, IvIg and other 12b 1.17 (0.73; 1.87) 10c 1.43 (0.99; 2.07)
Other 45d 0.87 (0.62; 1.21) 38e 0.90 (0.60; 1.33)
Primary recurrent miscarriage
Age at rst pregnancy after referral 290 0.98 (0.96; 1.00) 208 0.99 (0.97; 1.01)
Prior miscarriage 0.88 (0.78; 0.99) 0.79 (0.64; 0.97)
Prior non-visualized pregnancy loss 0.91 (0.82; 1.02) 0.88 (0.75; 1.04)
Prior biochemical pregnancy loss 0.90 (0.79; 1.02) 0.84 (0.70; 1.02)
Prior failed pregnancy of unknown location 0.96 (0.86; 1.08) 0.95 (0.78; 1.17)
Tender loving care 224 1 (reference) 159 1.00 (reference)
IvIg alone 45 1.39 (1.10; 1.76) 34 1.50 (1.15; 1.96)
IvIg and other 10f 1.40 (0.93; 2.12) 8g 1.76 (1.35; 2.31)
Other 11h 0.83 (0.43; 1.60) 7i 0.89 (0.39; 2.03)
Secondary recurrent miscarriage
Age at rst pregnancy after referral 209 0.97 (0.95; 1.00) 160 0.97 (0.94; 1.01)
Prior miscarriage 0.84 (0.75; 0.94) 0.82 (0.72; 0.94)
Prior non-visualized pregnancy loss 0.86 (0.76; 0.97) 0.86 (0.75; 0.99)
Prior biochemical pregnancy loss 0.89 (0.79; 1.01) 0.91 (0.79; 1.04)
Prior failed pregnancy of unknown location 0.79 (0.65; 0.97) 0.77 (0.57; 1.04)
Tender loving care 120 1 (reference) 84 1.00 (reference)
IvIg alone 53 1.15 (0.89; 1.48) 43 1.18 (0.87; 1.60)
IvIg and other 2j 0 (0.00; 0.00) 2k 0.00 (0.00; 0.00)
Other 34l 0.85 (0.57; 1.28) 31m 0.84 (0.53; 1.33)

IvIg, intravenous immunoglobulin.

a
Non-visualized pregnancy loss: combines biochemical pregnancy losses and failed pregnancy of unkown location (PUL).
b
Includes: IvIg+progesterone (n 3); IvIg+heparin (n 1); IvIg+donor lymphocytes (n 1); IvIg+prednisone (n 6); and IvIg+prednisone+progesterone (n 1).
c
Includes: IvIg+lymphocytes (n 1); IvIg+prednisone (n 5); IvIg+progesterone (n 3); and IvIg+progesterone+prednisone (n 1).
d
Includes: heparin (n 5), donor lymphocytes (n 4), prednisone (n 2), progesterone (n 7) and participants in a double blind randomized controlled trial of IvIg versus placebo (n 27).
e
Includes: donor lymphocyte (n 4); progesterone (n 5), heparin (n 4) and participants in a double blinded randomized trial of IvIg versus placebo (n 25).
f
Includes: IvIg+donor lymphocytes (n 1); IvIg+heparin (n 1); IvIg+prednisone (n 6); IvIg+prednisone+progesterone (n 1); and IvIg+progesterone (n 1).
g
Includes: IvIg+donor lymphocytes (n 1); IvIg+prednisone (n 5); IvIg+progesterone (n 1); and IvIg+progesterone+prednisone (n 1).
h
Includes: donor lymphocytes (n 4); prednisone (n 2); heparin (n 3) and progesterone (n 2);
i
Includes: donor lymphocytes (n 4); progesterone (n 1); and heparin (n 2).
j
Includes: IvIg+progesterone (n 2).
k
Includes IvIg+progesterone (n 2)
l
Includes: heparin (n 2), progesterone (n 5) and participants in double blind randomized controlled trial of IvIg versus placebo (n 27).
m
Includes progesterone (n 4), heparin (n 2) and participants in a double blinded randomized trial of IvIg versus placebo (n 25).

accurate. This is substantiated by our data validation where the consist- distinguish between women with regular and irregular menstrual
ency between patient les and information given at rst consultation was cycles. In our study, we did not demonstrate a negative impact of high
99%. BMI on the chance of live birth.
Increased maternal age decreases the RR for live birth in the rst preg-
Life style factors nancy after referral. The RR described in this study is for each additional
High BMI has been reported to be prognostically negative in RM (Lashen year, and as such aligns well with previously published studies (Brigham
et al., 2004; Lo et al., 2012). In these studies the authors do not et al., 1999; Lund et al., 2012).

Downloaded from https://academic.oup.com/humrep/article-abstract/29/5/931/675614/Non-visualized-pregnancy-losses-are-prognostically

by E-nformation user
on 03 October 2017
 Early pregnancy losses and recurrent miscarriage 937

Aetiology of non-visualized pregnancy losses Funding

It is probable that some of the non-visualized pregnancy losses are due to
No specic funding was sought for the study.
chromosome anomalies, as documented by several studies (Wilcox
et al., 1987; Zinaman et al., 1996; Wang et al., 2003) and reviewed by
Macklon et al. (2002). On the other hand, more of the non-visualized Conict of interest
pregnancy losses would have been classied as miscarriages if the
The authors declare that they have no conicts of interest.
women had been monitored as extensively in their rst pregnancies as
they are in the Danish RM Unit, where all patients are followed with ultra-
sound from 6 weeks gestation. References
We found that the frequency of surgically treated EPs in PRM womens re-
ASRM Practice Committee. Denitions of infertility and recurrent pregnancy
productive history was signicantly higher if they had no clinical miscarriages in loss: a committee opinion. Fertil Steril 2013;99:63.
their reproductive history compared with those with at least one conrmed Barnhart K, van Mello NM, Bourne T, Kirk E, Van CB, Bottomley C, Chung K,
intrauterine pregnancy loss. Therefore, we propose that at least some of Condous G, Goldstein S, Hajenius PJ et al. Pregnancy of unknown location:
their non-visualized pregnancy losses may be spontaneously resolved EPs. a consensus statement of nomenclature, denitions, and outcome. Fertil
Steril 2011;95:857 866.
Brigham SA, Conlon C, Farquharson RG. A longitudinal study of pregnancy
Conclusions outcome following idiopathic recurrent miscarriage. Hum Reprod 1999;
To our knowledge, this is the rst study to report on the occurrence and 14:2868 2871.
impact of non-visualized pregnancy losses in women with unexplained Deddens JA, Petersen MR. Approaches for estimating prevalence ratios.
Occup Environ Med 2008;65:501 506.
RM after spontaneous conception and 3 consecutive pregnancy losses.
Jauniaux E, Farquharson RG, Christiansen OB, Exalto N. Evidence-based
We demonstrate that non-visualized pregnancy losses are frequent.
guidelines for the investigation and medical treatment of recurrent
Each additional non-visualized pregnancy loss decreases the RR for live miscarriage. Hum Reprod 2006;21:2216 2222.
birth with 10%, which is the same impact conferred by a clinical miscar- Kirk E, Condous G, Bourne T. Pregnancies of unknown location. Best Pract
riage. For women with at least two clinical miscarriages the RR for live Res Clin Obstet Gynaecol 2009;23:493 499.
birth is decreased with almost 10% by each non-visualized pregnancy Lashen H, Fear K, Sturdee DW. Obesity is associated with increased risk of
loss and by almost 20% for each clinical miscarriage. The data and rst trimester and recurrent miscarriage: matched case-control study.
results support the inclusion of non-visualized pregnancy losses in deni- Hum Reprod 2004;19:1644 1646.
tions of RM. Further studies are needed to conrm or refute our ndings. Lo W, Rai R, Hameed A, Brailsford SR, Al-Ghamdi AA, Regan L. The effect of
body mass index on the outcome of pregnancy in women with recurrent
miscarriage. J Family Community Med 2012;19:167 171.
Supplementary data Lund M, Kamper-Jorgensen M, Nielsen HS, Lidegaard O, Andersen AM,
Christiansen OB. Prognosis for live birth in women with recurrent
Supplementary data are available at http://humrep.oxfordjournals.org/.
miscarriage: what is the best measure of success? Obstet Gynecol 2012;
119:37 43.
Acknowledgements Macklon NS, Geraedts JP, Fauser BC. Conception to ongoing pregnancy:
the black box of early pregnancy loss. Hum Reprod Update 2002;
We thank biostatistician Helle Hjmark Eriksen, Department of Hospital 8:333 343.
Affairs, Aalborg University Hospital, Aalborg, Denmark for statistical Regan L, Rai R, Backos M. The investigation and treatment of couples with
analysis in STATA. recurrent rst-trimester and second-trimester miscarriage. RCOG Green
Top Guideline 2011;17:1 17.
Saravelos SH, Li TC. Unexplained recurrent miscarriage: how can we explain
Authors roles it? Hum Reprod 2012;27:1882 1886.
Wang X, Chen C, Wang L, Chen D, Guang W, French J. Conception, early
A.M.K. collected data, entered data in a database, did quality control,
pregnancy loss, and time to clinical pregnancy: a population-based
performed statistical analyses in SAS 9.2 and wrote the paper.
prospective study. Fertil Steril 2003;79:577 584.
R.H.v.O., S.Q., R.G.F., M.S. and M.G. contributed to interpretation of Wilcox AJ, Baird DD, Weinberg CR, Armstrong EG, Musey PI, Wehmann RE,
data and critically revised the manuscript. O.B.C. initiated the study, par- Caneld RE. The use of biochemical assays in epidemiologic studies of
ticipated in data collection and contributed to interpretation of data and reproduction. Environ Health Perspect 1987;75:2935.
critically revised the manuscript. All co-authors approved the nal manu- Zinaman MJ, Clegg ED, Brown CC, OConnor J, Selevan SG. Estimates of
script before submission. human fertility and pregnancy loss. Fertil Steril 1996;65:503 509.

Downloaded from https://academic.oup.com/humrep/article-abstract/29/5/931/675614/Non-visualized-pregnancy-losses-are-prognostically

by E-nformation user
on 03 October 2017