doi: 10.1111/j.1365-3016.2008.00965.

587

Comparison of gestational age at birth based on last menstrual

period and ultrasound during the rst trimester
Caroline S. Hoffmana, Lynne C. Messerd, Pauline Mendolad, David A. Savitze, Amy H. Herringb,c and Katherine E. Hartmannf,g
a

Department of Epidemiology, bDepartment of Biostatistics and cCarolina Population Center, University of North Carolina at Chapel Hill, Chapel

Hill, NC, dHuman Studies Division, National Health and Environmental Effects Research Laboratory, US Environmental Protection Agency,
Research Triangle Park, NC, eDepartment of Community and Preventive Medicine, Mount Sinai School of Medicine, New York, NY, fDepartment
of Obstetrics and Gynecology, Vanderbilt University School of Medicine, and gInstitute for Medicine and Public Health, Vanderbilt University
Medical Center, Nashville, TN, USA

Summary
Correspondence:
Dr Caroline Hoffman,
National Institute of
Environmental Health
Sciences, Division of
Extramural Research and
Training, P.O. Box 12233,
Research Triangle Park,
NC 27709, USA.
E-mail:
dilworthch@niehs.nih.gov

Hoffman CS, Messer LC, Mendola P, Savitz DA, Herring AH, Hartmann KE. Comparison of gestational age at birth based on last menstrual period and ultrasound during
the rst trimester. Paediatric and Perinatal Epidemiology 2008; 22: 587596.
Reported last menstrual period (LMP) is commonly used to estimate gestational age
(GA) but may be unreliable. Ultrasound in the rst trimester is generally considered a
highly accurate method of pregnancy dating. The authors compared rst trimester
report of LMP and rst trimester ultrasound for estimating GA at birth and examined
whether disagreement between estimates varied by maternal and infant characteristics.
Analyses included 1867 singleton livebirths to women enrolled in a prospective pregnancy cohort. The authors computed the difference between LMP and ultrasound GA
estimates (GA difference) and examined the proportion of births within categories of
GA difference stratied by maternal and infant characteristics. The proportion of births
classied as preterm, term and post-term by pregnancy dating methods was also
examined.
LMP-based estimates were 0.8 days (standard deviation = 8.0, median = 0) longer on
average than ultrasound estimates. LMP classied more births as post-term than ultrasound (4.0% vs. 0.7%). GA difference was greater among young women, non-Hispanic
Black and Hispanic women, women of non-optimal body weight and mothers of
low-birthweight infants. Results indicate rst trimester report of LMP reasonably
approximates gestational age obtained from rst trimester ultrasound, but the degree
of discrepancy between estimates varies by important maternal characteristics.
Keywords: gestational age, measurement, accuracy, LMP, ultrasound estimates, maternal
age, ethnic group, preterm, post-term, bias.

Introduction
Self-report of last menstrual period (LMP) is often used
to estimate gestational age in clinical care and epidemiological research. However, reported LMP may be
unreliable.1 Women may not recall LMP date accurately2,3 or misinterpret early pregnancy bleeding as
normal menses.4 In addition, the convention of assigning a 14-day interval between menstruation and ovulation, implicit in calculations of gestational age, will be
inaccurate for women with irregular menstrual cycles
or delayed ovulation.5,6 This presents the dilemma of

whether reported LMP has acceptable precision for

research on pregnancy outcomes.
Ultrasonography before 20 weeks gestation is generally viewed as a more accurate method of estimating
gestational age than menstrual dating,7,8 with ultrasound assessment thought to have greater accuracy the
earlier it is conducted in pregnancy. Ultrasound assessment is limited by the implicit assumption that below
a certain gestational age all fetal size variability is due
to gestational age,9 which can lead to systematic underestimation of gestational age among pregnancies

Paediatric and Perinatal Epidemiology, 22, 587596. Journal Compilation 2008 Blackwell Publishing Ltd. No claim to original US government works

588

C. S. Hoffman et al.

exhibiting early growth restriction measured in later

scans.10 This is less of a concern for very early ultrasound scans given that variation in fetal size is minimal
during the rst trimester of pregnancy.11 In practice,
early ultrasounds may only be available from medical
records for a select group of individuals (i.e. those
for whom an early ultrasound is medically indicated),12,13 and generally have not been obtained
in large-scale studies.
Previous studies, which compared second trimester
ultrasound with LMP, found that ultrasound, on
average, resulted in lower (younger) gestational age
estimates than those derived from LMP7,1416 and that
the discrepancy between estimates was increased with
young maternal age, non-Hispanic Black and Hispanic
race/ethnicity, lower education, single marital status,
cigarette smoking, primiparity, non-optimal prepregnancy weight, and pre-pregnancy or gestational
diabetes.14,15,17,18 Ultrasound also resulted in a higher
proportion of preterm births and a lower proportion of
post-term births relative to LMP.10,14,15,1719 Systematic
inaccuracy of gestational age dating could distort estimated associations between maternal characteristics
and adverse pregnancy outcomes derived using gestational age (e.g. preterm birth) and result in outcome
misclassication.
We compared estimates of gestational age at birth
based on rst trimester report of LMP and very early
ultrasound to determine whether LMP collected early
in pregnancy could accurately date pregnancies and
identify predictors of increased discrepancy between
gestational age estimates. To the authors knowledge,
this is the rst study to compare rst trimester ultrasound estimates of gestational age, collected on all
study participants at a median age of 9 weeks completed gestation using a standardised approach, with
estimated gestational age based on self-reported
LMP among a group of spontaneously conceived
pregnancies.

Methods
Study design and population
Right from the Start (RFTS) is an ongoing, communitybased prospective cohort study. This analysis includes
participants in RFTS Phase I, conducted from 2000 to
2004. Women who were early in pregnancy (12
weeks gestation) or planning to become pregnant
were recruited from private and public prenatal care

clinics and the community at large (e.g. using information posters and brochures, advertisements placed in
community publications, targeted mailings) and followed until the end of pregnancy. Additional details of
site selection and recruitment are described elsewhere.20 The Institutional Review Boards at the University of North Carolina, University of Tennessee and the
University of Texas approved the study protocols; participants gave informed consent.
Pregnant women were enrolled after providing an
LMP date and date of their rst positive pregnancy test
(n = 2514). Women trying to become pregnant were
pre-enrolled and followed for up to 6 months with
free pregnancy test kits provided; those who conceived
were formally enrolled (n = 252). In all, 2766 pregnancies were enrolled and available for assessment. We
excluded 347 (12.5%) pregnancies that ended in a loss,
91 (3.3%) pregnancies lost to follow-up, 8 (0.3%) multigestational pregnancies and 16 (0.6%) repeat livebirths
to a RFTS participant. Additional exclusions were
made for pregnancies with no or incomplete baseline
interview (n = 259; 9.4%), incomplete LMP date (n = 18;
0.6%), no ultrasound data (n = 158; 5.7%) and missing
infant birth date (n = 2; 0.07%). This analysis includes
1867 singleton livebirths with complete baseline interview, LMP (month and day) and ultrasound data.
Women retained for analysis were similar to all
enrolled participants with respect to maternal age, estimated gestational age at enrolment and parity, but
were slightly more likely to be White and to have completed 16 years of education.

Data collection
Information about LMP was obtained during screening
and enrolment telephone interviews, typically at 8
weeks gestation for women pregnant at initial screening and 5 weeks gestation for pre-enrolled women.
Women were informed they might need a calendar (or
diary) to answer certain questions and were invited to
obtain a calendar for the interview. Women were asked
What was the rst day of your last menstrual period?
If necessary, they were prompted to think of an event
that happened around the same time as their last
period (e.g. a holiday, vacation or weekend) to facilitate
recall. Additional maternal characteristics also were
collected by telephone interview.
First trimester endovaginal ultrasounds were conducted by clinical sonographers who were required to
have American Registry of Diagnostic Medical Sonog-

Paediatric and Perinatal Epidemiology, 22, 587596. Journal Compilation 2008 Blackwell Publishing Ltd. No claim to original US government works

Gestational age based on LMP vs. ultrasound

raphers (ARDMS, Rockville, MD, USA) certication,
use state-of-the-art equipment as assessed by a study
investigator (KH), conduct and document manufacturer recommended machine calibration, and have 3 or
more years of experience in pelvic and obstetric diagnostic sonography. In addition, sonographers participated in study-specic training. Measurements of the
gestational sac, yolk sac, fetal pole and fetal heart rate
were fully comparable with those required for clinical
pregnancy dating. A still image was reviewed by
trained staff prior to entry of ultrasound data, and a
20% quality sample was reviewed by a clinician skilled
in rst trimester sonography (KH). Mean gestational
age at ultrasound performance (according to selfreported LMP) was 9 weeks (median = 8 weeks,
range = 417 weeks).
Multiple data sources were used to obtain and
conrm livebirth outcomes: birth and fetal death
records, hospital discharge summaries, prenatal care
records and participant self-report. For this analysis,
pregnancy outcome information was documented
using medical records for 42.9% of livebirths, vital
records for 56.8% and participant self-report for only
0.4%.

Statistical analysis
We compared the distributions of gestational age at
birth (in days) from LMP (date of birth date of LMP)
and ultrasound [date of birth - (date of ultrasound
estimated clinical age at assessment)]. Like the LMPbased estimate, the ultrasound estimated clinical age
includes an assumed 14-day interval between menses
and conception, rendering the two estimated methods
comparable.
The difference in days between the LMP and ultrasound estimates was calculated (LMP estimate ultrasound estimate, hereafter referred to as gestational age

589

difference), and analysed as a continuous variable and

categorised into ve groups (<-14, -14 to -8, -7 to +7,
+8 to +14 and >+14 days). These categories were chosen
because they represent cut-points of discrepancy at
which a clinician would typically replace an estimate
of gestational age derived from the LMP with the ultrasound estimate (+/-7 days for rst trimester ultrasound and +/-14 days for second trimester
ultrasound). Positive values indicate that the LMPbased gestational age estimate is older than the
ultrasound-based estimate, and negative values indicate that the LMP-based gestational age estimate is
younger than the ultrasound-based estimate. Mean
and standard deviation values for the gestational age
difference and the proportion of births within categories of gestational age difference were calculated
within and compared across strata of maternal and
infant characteristics. Women were also stratied by
self-reported certainty of LMP date (very sure, pretty
sure, somewhat uncertain or very uncertain) and
pregnancy status (trying to conceive and already
pregnant) at initial screening. The already pregnant
women were further subdivided into three groups:
women who had not used contraceptives in the past
year or stopped using contraceptives to become pregnant, women who became pregnant during a gap or
break in contraceptive use, and women who got pregnant while using contraception.
Many of the maternal and infant characteristics
examined in this analysis are associated (e.g. young
mothers are also more likely to be single and less educated). Therefore, we also constructed unadjusted and
adjusted logistic regression models for the log odds of
a gestational age difference 8 days vs. +/-7 days and
a gestational age difference -8 days vs. +/-7 days to
determine whether associations indicated by nonparametric analysis and unadjusted odds ratios (OR)
remained after adjustment for other variables. All

Figure 1. Distribution of estimated

gestational age at birth by last menstrual
period (LMP) and ultrasound (US) (left
panel) and the distribution of number of
days difference between LMP and US
estimates (right panel) among livebirths
(n = 1867) to participants in the Right From
the Start study, 200004.

Paediatric and Perinatal Epidemiology, 22, 587596. Journal Compilation 2008 Blackwell Publishing Ltd. No claim to original US government works

590

C. S. Hoffman et al.

variables found to be associated with gestational age

difference in unadjusted models [maternal age,
race/ethnicity, educational level, marital status, prepregnancy body mass index (BMI), certainty of LMP
and infant birthweight] were included in the fully
adjusted model.
Kappa statistics and associated 95% condence intervals [CI] were calculated to assess agreement between
dating methods beyond chance for preterm, term and
post-term classication. We also calculated the proportion of births discordantly classied as preterm,
term or post-term by LMP and ultrasound. Analyses
were performed using Stata version 9 (College Station,
TX).

Results
The median estimated gestational age at birth was 276
days (39 weeks gestation) for both dating methods
(Fig. 1, left panel). Compared with LMP, ultrasoundderived gestational age was slightly shorter (mean of
273 vs. 274 days), less variable (standard deviation of
14 vs. 16 days), and had a narrower range (214307
days vs. 207315 days). A similar proportion of births
were classied as preterm (<37 weeks gestation) by
LMP than by ultrasound (proportions and associated
95% CIs were 10.0 [8.7, 11.5] vs. 8.9 [7.6, 10.3] respectively). The proportion of births classied as postterm (>41 weeks gestation) was notably higher
by LMP dating (4.0 [3.2, 5.0] vs. 0.7 [0.4, 1.3] %
respectively).
Gestational age difference was within +/-7 days for
80.8% of women and +/-30 days for 98.5% (Fig. 1,
right panel). The mean difference was 0.8 days
(SD = 8.0, median = 0 days) and ranged from -70 to 38
days. LMP-based estimates were somewhat more
likely to be higher (older) than ultrasound-based estimates when discrepant (gestational age difference was
zero for 10.5% of women, positive for 47.8% and negative for 41.7%).
Women <20 years old were more likely to have
both positive and negative differences (+8 days or
-8 days) than other age groups, while women 30
years old were less likely to have a positive gestational age difference (Table 1). Low education level
(completed high school or less), single marital status
and non-optimal BMI (<18.9 or >29.0 kg/m2) were
also associated with overall increased discrepancy.
Non-Hispanic Black women were more likely to have

a negative difference (-8 days) than non-Hispanic

White women. Conversely, women of Hispanic and
other ethnicity were more likely to have a positive
difference (+8 days) than non-Hispanic White
women.
An inverse relationship was observed for gestational
age difference and self-reported certainty of LMP
date, particularly for a positive difference >+14 days
among women who were only somewhat sure or
very uncertain (Table 1). Women who were planning
a pregnancy at initial screening were more likely to
have a positive difference and much less likely to have
a negative difference than women already pregnant.
Among those already pregnant, women who became
pregnant during a gap in contraceptive use or owing to
contraceptive failure had increased discrepancy in
both directions.
Low-birthweight infants (<2500 g at birth) and
infants weighing 25003499 g at birth were more likely
to have a positive difference of +8 to +14 days than
larger infants (Table 1). The degree of discrepancy
between estimates did not vary by maternal smoking,
parity or infant gender (results not shown). Exclusion
of women with a gestational age difference <-30 or >30
days did not affect the pattern of associations found
(results not shown).
Estimated ORs for a gestational age difference 8
days vs. +/-7 days and a gestational age difference
-8 days vs. +/-7 days are presented in Table 2.
Results of the unadjusted regression analyses
agree with those presented above. Most maternal
characteristics were no longer statistically signicant
(P-value > 0.05) after adjustment for other variables
(Table 2). However, the magnitude of the ORs for most
of these characteristics indicated a similar pattern of
associations with gestational age difference, with the
exception of single marital status, low education level
and Hispanic race/ethnicity, which were no longer
associated with gestational age difference after adjustment (i.e. ORs were attenuated towards the null after
adjustment).
Classication of births as preterm, term and postterm was concordant for 91.2% of all births (Fig. 2). The
Kappa coefcient for classication into these three categories was 0.59 [95% CI 0.54, 0.65], indicating moderate agreement.21 Agreement for classifying livebirths as
preterm (<37 vs. 37 weeks) was higher (Kappa coefcient = 0.75 [0.70, 0.80]). Agreement for classifying
livebirths as post-term ( 41 vs. >41 weeks) was very
low (Kappa coefcient = 0.05 [-0.02, 0.13]).

Paediatric and Perinatal Epidemiology, 22, 587596. Journal Compilation 2008 Blackwell Publishing Ltd. No claim to original US government works

Gestational age based on LMP vs. ultrasound

591

Table 1. Difference in LMP and ultrasound-based estimates of gestational age at birth by selected maternal and infant characteristics,
livebirths (n = 1867) to participants in the RFTS study, 200004
GA difference (in days): LMP US estimate

All participants
Maternal age (years)
<20
2024
2529
3034
35

Mean (SD)

<-14

-14 to -8

+8 to +14

>+14

1867

0.8 (8.0)

2.0%

4.5%

80.8%

8.6%

4.1%

P-valuea

<0.001

107
426
610
528
196

1.7
0.5
1.5
0.6
-0.6

(12.2)
(9.8)
(7.4)
(6.4)
(6)

2.8%
2.8%
1.5%
2.3%
1.0%

10.3%
5.6%
3.8%
2.8%
5.6%

65.4%
75.6%
81.2%
85.8
85.7%

12.2%
11.3%
8.7%
6.6%
6.1%

9.4%
4.7%
4.9%
2.5%
1.5%

Race/ethnicity
Non-Hispanic White
Non-Hispanic Black
Hispanic
Other
Missing

1109
520
176
59
3

1.2
-0.5
1.3
2.5

(6.5)
(9.7)
(9.3)
(10.1)

1.3%
3.3%
3.4%
1.7%

3.4%
7.7%
3.4%
0%

84.1%
76.5%
75.0%
72.9%

8.3%
7.7%
11.9%
13.6%

2.9%
4.8%
6.3%
11.9%

Education level
High school or less
Some college
4+ years of college

496
417
954

0.8 (10.8)
0.6 (7.1)
0.9 (6.5)

3.2%
1.9%
1.5%

5.7%
4.8%
3.8%

73.2%
80.6%
84.8%

11.9%
9.8%
6.4%

6.1%
2.9%
3.6%

<0.001

1298
568
1

1.1 (6.9)
0.1 (10.1)

1.6%
3.0%

3.4%
7.0%

83.4%
74.8%

8.0%
9.9%

3.5%
5.3%

<0.001

192
956
253
418
48

1.5
0.8
0.2
0.8

(7.6)
(7.5)
(8.0)
(9.0)

0.5%
1.4%
3.6%
2.9%

6.8%
4.0%
4.0%
5.0%

77.1%
84.0%
80.6%
76.6%

9.4%
7.2%
8.3%
10.8%

6.3%
3.5%
3.6%
4.8%

0.026

Certainty of LMP date

Very sure
Pretty sure
Somewhat sure
Very uncertain
Missing

1275
367
137
48
40

0.5
0.6
3.1
2.1

(7.2)
(8.7)
(11)
(12.9)

1.7%
2.7%
2.9%
4.2%

3.7%
6.8%
5.1%
8.3%

84.2%
76.6%
66.4%
56.3%

7.8%
9.3%
13.1%
14.6%

2.6%
4.6%
12.4%
16.7%

0.014b

Pregnancy status at screening

Trying to conceive
Pregnantc
No birth control use
Gap/break in birth control use
Birth control failure
Missing

153
1714
1236
323
147
8

3
0.6
-0.3
1.4
0.6

(6.8)
(8.1)
(8.3)
(11.4)
(8.2)

0%
2.2%
1.7%
4.0%
2.0%

0.7%
4.8%
3.4%
9.3%
6.8%

83.7%
80.5%
82.6%
74.9%
76.2%

9.8%
8.5%
9.3%
6.2%
7.5%

5.9%
3.9%
3%
5.6%
7.5%

<0.001

104
959
557
245
2

0.9
1.0
0.6
0.2

(10.0)
(8.4)
(7.2)
(7.2)

4.8%
2.2%
1.4%
1.6%

2.9%
5.4%
4.3%
2.0%

74.0%
77.9%
84.2%
86.9%

12.5%
9.9%
6.6%
6.5%

5.8%
4.6%
3.4%
2.9%

0.022

Marital status
Married
Not married
Missing
Pre-pregnancy BMI (kg/m2)
<19.8
19.825.9
26.029.0
>29.0
Missing

Infant birthweight (g)

<2500
25003499
35003999
4000
Missing

0.024b

a
P-value corresponds to two-sided Pearson chi-square test for independence of maternal/infant characteristics and categories of GA difference
unless otherwise specied.
b
P-value corresponds to two-sided Fishers Exact test for independence of maternal/infant characteristics and categories of GA difference.
c
Participants that were pregnant at enrolment were further subdivided into three categories: those who reported they had not used contraceptives
in the past year or had stopped using contraceptives to become pregnant, those who reported they became pregnant during a gap or break in
contraceptive use, and those who reported they became pregnant owing to contraceptive failure.
BMI, body mass index; GA, gestational age; LMP, last menstrual period; RFTS, Right from the Start; SD, standard deviation; US, ultrasound.

Paediatric and Perinatal Epidemiology, 22, 587596. Journal Compilation 2008 Blackwell Publishing Ltd. No claim to original US government works

592

C. S. Hoffman et al.

Table 2. Odds of having a difference of more than 7 days between LMP and US-based estimates of gestational age at birth by selected
maternal and infant characteristics: livebirths to participants in the RFTS study, 200004
GA difference +8 vs. -7 to +7a
+8
n
Maternal age (years)
<20
23
2024
68
2529
83
3034
48
35
15
P-value
Race/ethnicity
Non-Hispanic White
124
Non-Hispanic Black
65
Hispanic
32
15
Otherd
Missing
1
P-value
Education level
High school or less
89
Some college
53
4+ years of college
95
P-value
Marital status
Married
150
Not married
86
Missing
1
P-value
Pre-pregnancy BMI (kg/m2)
<19.8
30
19.825.9
102
26.029.0
30
>29.0
65
Missing
10
P-value
Certainty of LMP date
Very sure
133
Pretty sure
51
Somewhat sure
35
Very uncertain
15
Missing
3
P-value
Infant birthweight (g)
<2500
19
25003499
139
35003999
56
4000
23
Missing
0
P-value

GA difference -8 vs. -7 to +7b

+/-7
n

Unadjusted
OR [95% CI]

Adjustedc
OR [95% CI]

-8
n

+/-7
n

Unadjusted
OR [95% CI]

Adjustedc
OR [95% CI]

70
322
495
453
168

2.0 [1.2, 3.3]

1.3 [0.9, 1.8]
1.0 Reference
0.6 [0.4, 0.9]
0.5 [0.3, 0.9]
<0.001

1.5 [0.8, 2.8]

1.0 [0.7, 1.6]
1.0 Reference
0.7 [0.5, 1.1]
0.5 [0.3, 1.0]
0.084

14
36
32
27
13

70
322
495
453
168

3.1 [1.6, 6.1]

1.7 [1.1, 2.8]
1.0 Reference
0.9 [0.5, 1.6]
1.2 [0.6, 2.3]
0.002

1.7 [0.7, 3.8]

1.3 [0.7, 2.3]
1.0 Reference
1.1 [0.6, 1.9]
1.3 [0.7, 2.7]
0.762

933
398
132
43
2

1.0 Reference
1.2 [0.9, 1.7]
1.8 [1.2, 2.8]
2.6 [1.4, 4.9]

1.0 Reference
0.8 [0.6, 1.2]
1.2 [0.7, 1.9]
2.3 [1.2, 4.5]

52
57
12
1
0

933
398
132
43
2

1.0 Reference
2.6 [1.7, 3.8]
1.6 [0.8, 3.1]

1.0 Reference
1.8 [1.1, 2.9]
1.4 [0.7, 2.9]

0.002

0.033

<0.001

0.057

363
336
809

1.6 [1.1, 2.3]

1.0 Reference
0.7 [0.5, 1.1]
<0.001

1.3 [0.8, 1.9]

1.0 Reference
1.0 [0.7, 1.6]
0.494

44
28
50

363
336
809

1.5 [0.9, 2.4]

1.0 Reference
0.7 [0.5, 1.2]
0.008

1.1 [0.6, 1.9]

1.0 Reference
1.1 [0.6, 1.9]
0.930

1083
425
0

1.0 Reference
1.5 [1.1, 1.9]

1.0 Reference
1.0 [0.7, 1.5]

65
57
0

1083
425
0

1.0 Reference
2.2 [1.5, 3.2]

1.0 Reference
1.3 [0.8, 2.3]

0.010

0.995

<0.001

0.275

1.6 [1.0, 2.5]

1.0 Reference
1.2 [0.7, 1.8]
1.6 [1.1, 2.2]

1.4 [0.9, 2.2]

1.0 Reference
1.2 [0.8, 1.9]
1.6 [1.1, 2.3]

1.5 [0.8, 2.8]

1.0 Reference
1.5 [0.8, 2.5]
1.6 [1.0, 2.6]

1.5 [0.8, 2.8]

1.0 Reference
1.4 [0.8, 2.5]
1.4 [0.8, 2.2]

0.025

0.077

0.156

0.413

1.0 Reference
1.5 [1.0, 2.1]
3.1 [2.0, 4.8]
4.5 [2.3, 8.6]

1.0 Reference
1.5 [1.0, 2.1]
2.9 [1.8, 4.6]
3.3 [1.6, 6.9]

1.0 Reference
1.9 [1.3, 3.0]
1.9 [1.0, 3.7]
3.5 [1.4, 8.6]

1.0 Reference
1.9 [1.2, 3.0]
1.8 [0.9, 3.6]
2.2 [0.7, 6.6]

<0.001

<0.001

0.002

0.018

2.1 [1.2, 3.7]

1.6 [1.1, 2.2]
1.0 Reference
0.9 [0.5, 1.5]

2.0 [1.1, 3.8]

1.6 [1.1, 2.2]
1.0 Reference
1.0 [0.6, 1.6]

1.5 [0.7, 3.4]

1.4 [0.9, 2.2]
1.0 Reference
0.6 [0.3, 1.3]

1.1 [0.5, 2.7]

1.3 [0.8, 2.0]
1.0 Reference
0.7 [0.3, 1.5]

0.005

0.014

0.065

0.382

148
803
204
320
33

1073
281
91
27
36

77
747
469
213
2

14
51
19
33
5

69
35
11
6
1

8
73
32
9
0

148
803
204
320
33

1073
281
91
27
36

77
747
469
213
2

122 participants with GA difference -8 excluded from models.

237 participants with a GA difference +8 excluded from models.
c
Adjusted for all other variables in the table.
d
Number of subjects with Other race/ethnicity and a GA difference -8 was too small to include in the regression model, so subjects
were dropped from the unadjusted and adjusted models.
BMI, body mass index; CI, condence interval; GA, gestational age; LMP, last menstrual period; OR, odds ratio; RFTS, Right from the Start;
SD, standard deviation; US, ultrasound.
a

Paediatric and Perinatal Epidemiology, 22, 587596. Journal Compilation 2008 Blackwell Publishing Ltd. No claim to original US government works

Gestational age based on LMP vs. ultrasound

Figure 2. Scatter plot of estimated gestational age at birth

derived from last menstrual period (LMP) by estimated gestational age at birth derived from ultrasound among livebirths
(n = 1867) to participants in the Right From the Start study, 2000
04. Black line, ruled line at cut-point for dening preterm birth
(259 days); Dashed line, ruled line at cut-point for dening postterm birth (294 days).

Discussion
Findings from this study suggest that early report of
LMP results in similar gestational age dating as that
obtained from early ultrasound. Over 90% of women
were concordantly classied as preterm, term or postterm by LMP and ultrasound. However, the LMP estimate of gestational age tended to be greater than
gestational age based upon ultrasound when discrepant (by approximately 1 day, on average) and resulted
in a higher proportion of births classied as post-term.
In addition, the degree of difference between LMP and
ultrasound-based estimates varied by important maternal characteristics.
No feasible gold standard currently exists for estimating the gestational age of spontaneously conceived
pregnancies in large-scale epidemiological studies.
From a clinical perspective, rst trimester crown-rump
measurement is considered the most accurate method
of dating pregnancies.13 The validity of rst trimester
ultrasound assessment has been empirically demonstrated in studies of women undergoing in vitro fertilization, which have found high concordance between
known time of embryo transfer (oocyte retrieval + 14
days) and gestational age from rst trimester ultra-

593

sound, with mean differences in the range of -1 to +2

days.11,22,23 The documented validity of rst trimester
ultrasound encourages our condence in it as a
method of pregnancy dating. Furthermore, to minimise the potential for measurement error in our study,
ultrasounds were conducted using a standardised
approach that required sonographers to be certied
and have at least 3 years experience, use state-ofthe-art equipment that was regularly calibrated, and
undergo study-specic training. Therefore, we suspect
that the majority of discrepancies between LMP and
ultrasound-based estimates of gestational age over 7
days in our study are likely to be due to errors in
menstrual dating.
Gestational age miscalculation can arise from participant uncertainty in LMP, as evidenced by less discrepancy in gestational age estimates among women who
indicated they were very sure of their LMP compared
with women who were less sure. We also found that
the gestational age difference was lower among
women who were planning a pregnancy and presumably tracking their menstrual cycles closely, a nding
demonstrated previously.14 Greater overall discrepancy
between LMP and ultrasound estimates of gestational
age was also found among women of young maternal
age, lower education, single marital status and nonoptimal body weight, consistent with previous studies.14,15,17,24 Women with these characteristics may have
greater difculty recalling LMP in general,24 which
would contribute to random error in menstrual dating
and account for increases in both positive and negative
discrepancies.
Gestational age miscalculation also results from systematic error. Reported LMP underestimated
ultrasound-based gestational age by 8+ days for a
greater proportion of young women (<20 years) than
women of older ages. Early vaginal bleeding, often
confused with menses, is more prevalent among
young mothers.25 This biological phenomenon could
cause young women to systematically misrepresent
their LMP and explain the apparent mismatch between
the two dating methods. Furthermore, delayed ovulation, which is more common than early ovulation26 and
believed to be more prevalent among young women
and those of non-optimal body weight,27 could explain
the systematic overestimation of gestational age based
on LMP among these women.
We found that Hispanic women, who are predominately Mexican American in our study population,
were more likely to have a positive gestational age

Paediatric and Perinatal Epidemiology, 22, 587596. Journal Compilation 2008 Blackwell Publishing Ltd. No claim to original US government works

594

C. S. Hoffman et al.

difference than non-Hispanic White women, although

this association was attenuated after adjustment for
other maternal characteristics in the fully adjusted
model and our ndings are based on a very small,
non-random sample of Mexican American infants
(n = 176). A previous study has also reported greater
inconsistency between LMP-based and ultrasoundbased estimates of gestational age among Hispanic
women (both in a positive and negative direction).17
Additional research is needed to address whether gestation based on date of LMP is a potential source of bias
in studies of women with Hispanic ethnicity.
Among non-Hispanic Black women, LMP-based
gestational age was generally younger than the gestational age resulting from early ultrasound. One previous study also has reported a greater discrepancy
between LMP and ultrasound-based estimates of gestational age, but non-Hispanic Black women were
more likely than non-Hispanic white to have a difference in a positive or negative direction in that study.17
Given the persistent racial disparity in all adverse birth
outcomes between White and Black non-Hispanic
women, this could be an important nding from our
study. If, on average, Black womens reported LMP
systematically underestimates their infants actual
developmental age, as our study suggests, this may
explain some portion of the racial disparity in birth
outcomes. Verication in a representative sample is
needed.
We found an increased gestational age difference
among infants weighing <3500 g, predominately in a
positive direction. This is consistent with previous
studies based on second trimester ultrasound,15,28,29 and
may be due to systematic underestimation of gestational age by ultrasound dating among fetuses that
exhibit early growth restriction and remain small at
birth. However, growth restriction at 8 to 9 weeks
gestation, when ultrasounds predominately were conducted for this study, has not been previously reported.
Further research is needed to address whether rst
trimester growth restriction is a potential source of bias
in very early ultrasound assessments of gestational age,
or conversely, whether women who misreport their
LMP also are more likely to deliver a lowerbirthweight infant.
Unlike previous studies, in which ultrasounds were
predominantly based on second trimester ultrasounds,
we did not nd that maternal smoking and infant sex
were associated with increased positive gestational age
difference.9,14,15 Because smoking and female sex are

both associated with smaller fetal size, researchers

have attributed the differences in gestational age estimates associated with smoking and infant sex to errors
in ultrasound assessment, as ultrasound assessment
is based on the implicit assumption that there is
no within-gestational age variability in fetal size.9
However, previous studies have not found a discernable difference in crown-rump length measurements
by infant gender using ultrasound assessments conducted during the rst trimester30,31 suggesting that
variation in fetal size by gender does not occur until
later in pregnancy. This may explain why we did not
nd an association with gender in our study.
Two women in our study had gestational age difference values that were considerably lower than other
study participants (equal to -70 and -68). Gestational
age estimates for these women based on LMP vs. ultrasound were 29 vs. 39 weeks and 28 vs. 38 weeks
respectively. While birthweights of the infants born to
these women (3143 and 3371 g respectively) suggest
that the ultrasound estimate of gestational age at birth
is correct,32 both women reported they were very
sure of their self-reported LMP date. Therefore, we
decided not to exclude these outliers from the main
analysis. However, results were the same when we
re-ran analyses excluding these two women.
In our study, LMP was collected early in pregnancy
(by 12 weeks) and may not accurately reect errors in
menstrual dating that occur when LMP is recalled following a longer period (e.g. at infants birth).33 Therefore, our results may not be generalisable to studies
collecting LMP later in pregnancy. Furthermore, participants care providers were given a copy of the rst
trimester ultrasound report, which provided them
with an earlier (and hence, more accurate) estimate of
gestational age according to ultrasound than what is
usually available. Clinicians may have been more
willing to believe these ultrasounds and let women
misclassied as post-term by LMP only (i.e. not postterm by ultrasound) continue on in pregnancy. As a
result, our study may overstate the excess of births
classied as post-term by LMP compared with a more
general setting.
An additional consideration is the potential for
selection bias. Participants in the RFTS study represent
a group of women who knew they were pregnant in
the rst trimester and volunteered to participate in a
prospective pregnancy study, and thus may not be
comparable to other study samples or the general
population. In a previous report, we compared mater-

Paediatric and Perinatal Epidemiology, 22, 587596. Journal Compilation 2008 Blackwell Publishing Ltd. No claim to original US government works

Gestational age based on LMP vs. ultrasound

nal characteristics of pregnancies enrolled in the RFTS
study to all births identied from vital records in
the same geographical location over the same time
period.34 RFTS participants were similar to the general
population with respect to age but were more likely to
be highly educated (16 years of education), nonHispanic White and primiparous when compared with
the general population.
Given the early collection of LMP and ultrasound
and that our study population is highly selective, the
discrepancies between LMP and ultrasound estimates
of gestational age found in our study are likely to be
different from those in the general population. On the
other hand, rst trimester recall of LMP and rst
trimester ultrasounds are not relevant on women
who would not present in the same time frame, and
are probably unfeasible to collect in large-scale or
population-based studies. Thus, generalisability to a
more representative population is probably not a
worthwhile goal.
Gestational age remains an important measurement
in reproductive and perinatal research. Findings from
our study suggest that menstrual dating, based upon
LMP collected early in pregnancy, approximates reasonably to the gestational age obtained from rst trimester ultrasound, the most accurate pregnancy dating
method currently available. However, LMP tends to
overestimate gestational age, on average, and result in a
higher proportion of post-term births. Thus, LMP
should not be used to date pregnancies in studies
directed at examining factors associated with post-term
birth. The degree of difference between LMP and
ultrasound-based estimates also varies by important
maternal and infant characteristics (e.g. maternal race/
ethnicity), which should be taken into consideration
when designing and interpreting results from a study
examining the association between these characteristics
and pregnancy outcomes derived using gestational age.

Acknowledgements
This study was funded jointly by the Awwa Research
Foundation (AwwaRF) and the U.S. Environmental
Protection Agency (USEPA) under Cooperative
Agreement Nos. CR825625-01, CR827268-01 and
CR828216-01, a grant from the National Institute
of Environmental Health Sciences (P30ES10126), and
the NHEERL-DESE Cooperative Training grant in
Environmental Sciences Research (EPA CT8229471
and CR83323601).

595

Disclaimer
The information in this document has been funded, in
part, by the US Environmental Protection Agency. It
has been subjected to review by the National Health
and Environmental Effects Research Laboratory and
approved for publication. Approval does not signify
that the contents reect the views of the Agency, nor
does mention of trade names or commercial products
constitute endorsement or recommendation for use.

References
1 Berkowitz GS, Papiernik E. Epidemiology of preterm birth.
Epidemiologic Reviews 1993; 15:414443.
2 Wegienka G, Baird DD. A comparison of recalled date of
last menstrual period with prospectively recorded dates.
Journal of Womens Health 2005; 14:248252.
3 Waller DK, Spears WD, Gu Y, Cunningham GC. Assessing
number-specic error in the recall of onset of last menstrual
period. Paediatric and Perinatal Epidemiology 2000;
14:263267.
4 Gjessing HK, Skjaerven R, Wilcox AJ. Errors in gestational
age: evidence of bleeding early in pregnancy. American
Journal of Public Health 1999; 89:213218.
5 Waldenstrom U, Axelsson O, Nilsson S. Sonographic dating
of pregnancies among women with menstrual irregularities.
Acta Obstetricia et Gynecologica Scandinavica 1991; 70:1720.
6 Tunon K, Eik-Nes SH, Grottum P. A comparison between
ultrasound and a reliable last menstrual period as
predictors of the day of delivery in 15 000 examinations.
Ultrasound in Obstetrics and Gynecology 1996; 8:178185.
7 Kramer MS, McLean FH, Boyd ME, Usher RH. The validity
of gestational age estimation by menstrual dating in term,
preterm, and postterm gestations. JAMA 1988;
260:33063308.
8 Gardosi J, Geirsson RT. Routine ultrasound is the method of
choice for dating pregnancy. British Journal of Obstetrics and
Gynaecology 1998; 105:933936.
9 Henriksen TB, Wilcox AJ, Hedegaard M, Secher NJ. Bias in
studies of preterm and postterm delivery due to ultrasound
assessment of gestational age. Epidemiology 1995; 6:533537.
10 Olsen OE, Lie RT, Rosendahl K. Ultrasound estimates of
gestational age among perinatally demised: a
population-based study. Acta Obstetricia et Gynecologica
Scandinavica 2004; 83:149154.
11 Kalish RB, Thaler HT, Chasen ST, Gupta M, Berman SJ,
Rosenwaks Z, et al. First- and second-trimester ultrasound
assessment of gestational age. American Journal of Obstetrics
and Gynecology 2004; 191:975978.
12 Reuss ML, Hatch MC, Susser M. Early ultrasound dating of
pregnancy: selection and measurement biases. Journal of
Clinical Epidemiology 1995; 48:667674.
13 American College of Obstetricians and Gynecologists.
Ultrasonography in pregnancy. ACOG Practice Bulletin No.
58. Obstetrics and Gynecology 2004; 104:14491458.

Paediatric and Perinatal Epidemiology, 22, 587596. Journal Compilation 2008 Blackwell Publishing Ltd. No claim to original US government works

596

C. S. Hoffman et al.

14 Savitz DA, Terry JW, Dole N, Thorp JM, Siega-Riz AM,

Herring AH. Comparison of pregnancy dating by last
menstrual period, ultrasound scanning, and their
combination. American Journal of Obstetrics and Gynecology
2002; 187:16601666.
15 Morin I, Morin L, Zhang X, Platt RW, Blondel B, Breart G,
et al. Determinants and consequences of discrepancies in
menstrual and ultrasonographic gestational age estimates.
BJOG 2005; 112:145152.
16 Olesen AW, Westergaard JG, Thomsen SG, Olsen J.
Correlation between self-reported gestational age and
ultrasound measurements. Acta Obstetricia et Gynecologica
Scandinavica 2004; 83:10391043.
17 Dietz PM, England LJ, Callaghan WM, Pearl M, Wier ML,
Kharrazi M. A comparison of LMP-based and
ultrasound-based estimates of gestational age using linked
California livebirth and prenatal screening records.
Paediatric and Perinatal Epidemiology 2007; 21(Suppl. 2):6271.
18 Haglund B. Birthweight distributions by gestational age:
comparison of LMP-based and ultrasound-based estimates
of gestational age using data from the Swedish Birth
Registry. Paediatric and Perinatal Epidemiology 2007; 21(Suppl.
2):7278.
19 Lynch CD, Zhang J. The research implications of the
selection of a gestational age estimation method. Paediatric
and Perinatal Epidemiology 2007; 21(Suppl. 2):8696.
20 Promislow JH, Makarushka CM, Gorman JR, Howards PP,
Savitz DA, Hartmann KE. Recruitment for a
community-based study of early pregnancy: the Right from
the Start study. Paediatric and Perinatal Epidemiology 2004;
18:143152.
21 Stokes M, Davis CS, Koch GG. Observer agreement. In:
Categorical Data Analysis using the SAS System. 2nd edn.
Cary, NC: SAS Institute Inc, 2000; pp. 112115.
22 Sladkevicius P, Saltvedt S, Almstrom H, Kublickas M,
Grunewald C, Valentin L. Ultrasound dating at 1214 weeks
of gestation: a prospective cross-validation of established
dating formulae in in-vitro fertilized pregnancies.
Ultrasound in Obstetrics and Gynecology 2005; 26:504511.
23 Tunon K, Eik-Nes SH, Grottum P, Von During V, Kahn JA.
Gestational age in pregnancies conceived after in vitro
fertilization: a comparison between age assessed from
oocyte retrieval, crown-rump length and biparietal
diameter. Ultrasound in Obstetrics and Gynecology 2000;
15:4146.

24 Yang H, Kramer MS, Platt RW, Blondel B, Breart G, Morin I,

et al. How does early ultrasound scan estimation of
gestational age lead to higher rates of preterm birth?
American Journal of Obstetrics and Gynecology 2002;
186:433437.
25 Stevens-Simon C, Roghmann KJ, McAnarney ER. Early
vaginal bleeding, late prenatal care, and misdating in
adolescent pregnancies. Pediatrics 1991; 87:838840.
26 Baird DD, McConnaughey DR, Weinberg CR, Musey PI,
Collins DC, Kesner JS, et al. Application of a method for
estimating day of ovulation using urinary estrogen and
progesterone metabolites. Epidemiology 1995; 6:547550.
27 Harlow SD, Ephross SA. Epidemiology of menstruation and
its relevance to womens health. Epidemiologic Reviews 1995;
17:265286.
28 Tunon K, Eik-Nes SH, Grottum P. Fetal outcome when the
ultrasound estimate of the day of delivery is more than 14
days later than the last menstrual period estimate.
Ultrasound in Obstetrics and Gynecology 1999; 14:1722.
29 Nakling J, Backe B. Adverse obstetric outcome in fetuses
that are smaller than expected at second trimester routine
ultrasound examination. Acta Obstetricia et Gynecologica
Scandinavica 2002; 81:846851.
30 Tezuka N, Banzai M, Sato S, Saito H, Hiroi M. Sexual
difference in early fetal crown-rump length versus
gestational age in pregnancies arising from in vitro
fertilization. Gynecologic and Obstetric Investigation 1998;
45:151153.
31 Becker S, Ural S, Fehm T, Bienstock J. Fetal gender and
sonographic assessment of crown-rump length: implications
for multifetal pregnancy reduction. Ultrasound in Obstetrics
and Gynecology 2004; 24:399401.
32 Alexander GR, Himes JH, Kaufman RB, Mor J, Kogan M. A
United States national reference for fetal growth. Obstetrics
and Gynecology 1996; 87:163168.
33 Pearl M, Wier ML, Kharrazi M. Assessing the quality of last
menstrual period date on California birth records. Paediatric
and Perinatal Epidemiology 2007; 21(Suppl. 2):5061.
34 Savitz DA, Singer PC, Hartmann KE, Herring AH,
Weinberg HS, Makarushka C, et al. Drinking Water
Disinfection By-Products and Pregnancy Outcome. Final report
for AwwaRF Project #2579. Denver, CO: American Water
Works Association Research Foundation, 2005.

Paediatric and Perinatal Epidemiology, 22, 587596. Journal Compilation 2008 Blackwell Publishing Ltd. No claim to original US government works