62

A comparison of LMP-based and ultrasound-based estimates of

gestational age using linked California livebirth and prenatal
screening records
Patricia M. Dietza, Lucinda J. Englanda, William M. Callaghana, Michelle Pearlb, Megan L. Wierb and Martin Kharrazic
a

National Center for Chronic Disease Prevention and Health Promotion, Division of Reproductive Health, Centers for Disease Control and

Prevention, Atlanta, GA, bSequoia Foundation, La Jolla, and cCalifornia Department of Health Services, Genetic Disease Screening Program,
Richmond, CA, USA

Summary
Correspondence:
Patricia M. Dietz, 4770 Buford
Hwy MS K-22, Atlanta, GA
30341, USA.
E-mail: pad8@cdc.gov

Conicts of interest:
the authors have declared no
conicts of interest.

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.
Although early ultrasound (<20 weeks gestation) systematically underestimates the
gestational age of smaller fetuses by approximately 12 days, this bias is relatively small
compared with the large error introduced by last menstrual period (LMP) estimates of
gestation, as evidenced by the number of implausible birthweight-for-gestational age.
To characterise this misclassication, we compared gestational age estimates based on
LMP from California birth certicates with those based on early ultrasound from a
California linked Statewide Expanded Alpha-fetoprotein Screening Program (XAFP).
The nal sample comprised 165 908 women. Birthweight distributions were plotted by
gestational age; sensitivity and positive predictive value for preterm rates according to
LMP were calculated using ultrasound as the gold standard.
For gestational ages 2027 and 2831 weeks, the LMP-based birthweight distributions were bimodal, whereas the ultrasound-based distributions were unimodal, but
had long right tails. At 3236 weeks, the LMP distribution was wider, atter, and
shifted to the right, compared with the ultrasound distribution. LMP vs. ultrasound
estimates were, respectively, 8.7% vs. 7.9% preterm (<37 weeks), 81.2% vs. 91.0% term
(3741 weeks), and 10.1% vs. 1.1% post-term (42 weeks). The sensitivity of the LMPbased preterm birth estimate was 64.3%, and the positive predictive value was 58.7%.
Overall, 17.2% of the records had estimates with an absolute difference of >14 days.
The groups most likely to have inconsistent gestational age estimates included African
American and Hispanic women, younger and less-educated women, and those who
entered prenatal care after the second month of pregnancy. In conclusion, we found
substantial misclassication of LMP-based gestational age.
The 2003 revised US Standard Certicate of Live Birth includes a new gestational age
item, the obstetric estimate. It will be important to assess whether this estimate
addresses the problems presented by LMP-based gestational age.
Keywords: gestation, ultrasound estimate, LMP estimate, perterm rate, post-term rate.

Introduction
Problems with the accuracy of gestational age computed by last menstrual period (LMP) on birth certicates have been documented.15 Evidence of this

inaccuracy is illustrated by birthweight distributions

that are bimodal at gestational ages <32 weeks, with
the modal birthweight of the second peak consistent
with that of term infants.1,4 Inaccuracy of LMP-based

2007 Blackwell Publishing Ltd. No claim to original US government works. Paediatric and Perinatal Epidemiology, 21 (Suppl. 2), 6271

LMP and ultrasound estimates of gestational age

gestational age can be caused by biologically associated
errors in menstrual cycles and by human error in recall
or data entry.6,7 Inherent in estimating gestational age
with LMP is the assumption that all women have a
regular 28-day menstrual cycle and ovulate 14 days
after the rst day of their LMP. However, because
timing of ovulation varies, even with accurate recall
and data entry of the LMP, estimates of gestational age
based on LMP can be inaccurate. For example, one
study found that 10% of women had cycles <25 days
long, 12% were between 31 and 35 days, and 3% were
36 days or longer, while 5% were too irregular to say.8
Time from LMP to ovulation is more likely to be
longer, as opposed to shorter, than 14 days, resulting in
an overestimation of gestational age when using
LMP.9,10 Biologically associated error can also occur if
early bleeding in pregnancy is thought to be menstruation or if LMP is missing because of amenorrhoea.
Clinicians are well aware of the shortcomings of LMP,
and standard practice is to base gestational age estimates on early ultrasound (<20 weeks) or other factors
when LMP is uncertain. In addition, clinicians frequently substitute ultrasound-based gestational age
estimates for LMP-based estimates when the two
disagree.11
However, while early ultrasound has been established clinically as the gold standard, questions have
been raised as to its validity for use in research. One
common concern is that ultrasound may introduce
biases because it is based on fetal growth, and thus
could systematically result in the assignment of
incorrect lower gestational age estimates for smaller
infants.12,13 Recent studies have found that early
(<20 weeks) ultrasound-based gestational age formulas
are fairly accurate, with random errors of 10 days
[95% condence interval (CI)].14 In addition, fetuses
with characteristics associated with small fetal size,
such as rst births and female sex, were found to be
systematically dated 12 days younger.15,16 Another
large study of singleton pregnancies with ultrasound
examinations between 12 and 22 weeks found no evidence that growth-restricted fetuses were systematically classied incorrectly at lower gestational ages,
and that the discrepancy between the LMP-based
gestational age and the ultrasound-based gestational
age was primarily related to ovulation later than the
assumed 14 days.17 Thus, while early ultrasound may
systematically underestimate gestational age for
smaller fetuses by 12 days on average, this bias is
relatively small compared with the large magnitude of

63

error indicated by records with implausible

birthweight-for-gestational age based on LMP.4,9,18 Previous studies comparing LMP-based and ultrasoundbased gestational age have found high rates of
gestational age misclassication by LMP. However,
these studies have been limited to clinic- or hospitalbased samples,9,12,18 to women with reliable LMP
dates,12 or to studies outside the US.12,18 Therefore,
whether the ndings of these studies can be generalised to other populations is unknown.
We sought to better understand and characterise the
misclassication found with gestational age estimated
by using LMP from birth certicates. Because US
birth certicates do not include information on early
ultrasound, we compared gestational age estimates
based on LMP from California birth certicates with
gestational age estimates based on early ultrasound
(20 weeks gestation) from a population-based prenatal screening programme in which approximately 70%
of the States pregnant women participate. Unlike
previous studies, this inquiry beneted from a large
sample derived from the cohort of women who delivered in California in 2002.

Methods
The study population was dened as pregnant women
enrolled in the Statewide California Expanded
Alpha-fetoprotein Screening Program (XAFP) who
gave birth to a live singleton infant during 2002, and
who had an estimated gestational age based on ultrasound recorded on their XAFP screening form. The
XAFP is a triple marker screening programme offered
to all women entering prenatal care by 20 weeks gestation. When maternal blood is drawn for this screen,
the medical provider lls out a form dating the pregnancy based on LMP, physical examination, and/or
ultrasound, when available. Using SuperMatch 2001
software (SuperMATCH Concepts and Reference
Version 3.10, Vality Technology Incorporated, March
2001), a probabilistic method was employed to link
records from the XAFP and Statewide Newborn
Screening programmes and birth certicates using
mothers name, date of birth, social security number,
delivery date, XAFP accession date, telephone number,
street address, city and zip code. A conservative certainty cut-off was used to minimise false matches.
In 2002, there were 530 926 livebirths in California.
Of these, 327 218 livebirth records (62%) linked to an
XAFP record from the same pregnancy, with approxi-

2007 Blackwell Publishing Ltd. No claim to original US government works. Paediatric and Perinatal Epidemiology, 21 (Suppl. 2), 6271

64

P. M. Dietz et al.

Table 1. Maternal demographic and

pregnancy characteristics by study
eligibility and inclusion status, California
livebirths, 2002

Characteristic
Race/ethnicity
White
African American
Asian
Hispanic
Other
Age (years)
<20
2024
2534
35
Education (years)
<12
12
>12
Payment source (delivery)
Medi-Cal
Private
Uninsured
Other
Month prenatal care began
12
34
56
7
Parity
0
1
2
Infant birthweight (g)
<2500
2500
LMP-based gestational age (weeks)
<37
3741
4244
45

Eligible and included

(n = 165 908)
%

Ineligible or excluded
(n = 349 481)
%

31.8
5.4
7.8
48.3
6.7

30.2
6.1
6.8
51.0
5.8

7.4
21.0
58.3
13.2

10.9
24.6
46.9
17.6

24.9
28.1
47.0

31.1
29.0
39.9

36.1
62.2
1.1
0.6

46.7
47.8
2.9
2.5

76.1
22.1
NA
NA

65.7
24.3
6.4
3.6

40.6
32.8
26.5

38.6
31.3
30.1

4.9
95.1

5.0
95.0

8.7
81.2
8.0
2.1

9.0
84.2
5.6
1.2

LMP, last menstrual period; NA, not applicable.

Due to rounding or missing values totals may not add up to 100%.

mately 86% of XAFP records successfully linking to a

livebirth record. Failure to link records may have
resulted from data entry errors, pregnancies that did
not end in a livebirth, or women who moved out of
State before delivery. Among the linked records,
195 616 (59.8%) women had ultrasound reported on
the XAFP records. After excluding 3238 women with
multiple births, 192 378 women were eligible for the
study. Of these, we excluded records missing LMP on
the birth certicate (n = 26 249) or with gestational age
at birth of <20 weeks by either LMP (n = 206) or ultra-

sound (n = 30). The nal sample comprised 165 908

women (50.7% of livebirth records linked to an XAFP
record, 32.2% of livebirths in California in 2002,
Table 1).
LMP-based gestational age at delivery was calculated using LMP and date of birth from the birth certicate. Ultrasound-based gestational age at delivery
was calculated using the ultrasound-based estimate of
gestational age on the date the ultrasound was performed, and the date of delivery on the birth certicate.
We categorised the two gestational age variables into

2007 Blackwell Publishing Ltd. No claim to original US government works. Paediatric and Perinatal Epidemiology, 21 (Suppl. 2), 6271

LMP and ultrasound estimates of gestational age

ve groups based on completed weeks: 2027, 2831,
3236, 3741 and 42 weeks.
To explore predictors of inconsistent gestational age,
we obtained infant birthweight, race/ethnicity, mothers age, education, source of payment for delivery,
and month of entry into prenatal care from the birth
certicate.
We rst compared the birthweight distributions for
each gestational age group using LMP-based and
ultrasound-based gestational age estimates. We also
calculated the sensitivity and positive predictive value
of the LMP-based gestational age, using ultrasound as
the gold standard. We compared the mean birthweight
and whether the infant was placed in a neonatal intensive care unit (NICU) for estimates that were concordant and discordant for gestational age. For this
analysis only we divided the group of 2027 weeks into

65

two gestational age categories (2023 and 2427 weeks)

to more closely examine differences. The NICU variable was obtained from the Statewide Newborn
Screening programme database, and indicates whether
the infant was in a NICU at the time of specimen collection (median time between delivery and specimen
collection, 29 h). We compared the demographic characteristics of women with inconsistent ultrasound- and
LMP-based gestational age estimates. We dened
inconsistent as an absolute difference >14 days and
used this cut-off to identify gross errors in gestational
age. All demographic characteristics were entered into
a logistic regression model to assess the independent
effects of each risk factor on inconsistent estimates,
holding the other characteristics constant. Finally, we
calculated preterm delivery rates for LMP- and
ultrasound-based estimates overall, and by race/

40
30

Ultrasound

% of
20
births

LMP

10

Figure 1. Birthweight distribution of singleton

births delivered at 2027 weeks gestation
according to ultrasound (n = 733) and last
menstrual period (LMP) (n = 745).

0
400

1200 2200 3000

3800

Birthweight (g)

40
35
30
25
% of
20
births
15
10
5
0
400

Ultrasound
LMP

1200 2000

Figure 2. Birthweight distribution of singleton

births delivered at 2831 weeks gestation
according to ultrasound (n = 1091) and last
menstrual period (LMP) (n = 1235).

2800 3600 4400

Birthweight (g)

40
30
Ultrasound

% of 20
births

LMP

10

00
34
00
40
00
46
00

28

00

00

22

00

16

10

40

Birthweight (g)

Figure 3. Birthweight distribution of singleton

births delivered at 3236 weeks gestation
according to ultrasound (n = 11 410) and last
menstrual period (LMP) (n = 12 499).

2007 Blackwell Publishing Ltd. No claim to original US government works. Paediatric and Perinatal Epidemiology, 21 (Suppl. 2), 6271

66

P. M. Dietz et al.

ethnicity, age, parity, education, month of entry into

prenatal care, and infants sex.

Results
Women included in the sample differed from those not
included in that they were disproportionately aged
2534 years, more educated and less likely to have
Medi-Cal
(Californias
Medicaid
programme)
(Table 1). The women included were also more likely to
have begun prenatal care in the rst 2 months of pregnancy and were more likely to have delivered postterm (42 weeks gestation) based on LMP. The two

Table 2. Sensitivity and positive predictive

value of last menstrual period estimate of
gestational age using ultrasound estimates
as the gold standard, total study
population and by racial/ethnic groups

Gestational age (weeks)

All women
<37
2027
2831
3236
3741
42
White
<37
2027
2831
3236
3741
42
African American
<37
2027
2831
3236
3741
42
Hispanic
<37
2027
2831
3236
3741
42
Asian
<37
2027
2831
3236
3741
42

groups were similar in racial/ethnic and low birthweight rates.

Figures 13 present the birthweight distribution
by LMP- and ultrasound-based gestational age. For
LMP-based gestational ages 2027 weeks (Fig. 1) and
2831 weeks (Fig. 2), the birthweight distribution is
bimodal, whereas the distribution based on ultrasound
gestational age is not, but it has a long right tail. For
LMP-based gestational age 3236 weeks (Fig. 3), the
birthweight distribution is wider, atter, and shifted
to the right compared with the ultrasound-based
distribution. For both LMP- and ultrasound-based
gestational ages 3741 weeks (gure not shown),

Sensitivity
% [95% CI]

Positive predictive value

% [95% CI]

64.3
76.9
60.4
57.6
85.6
33.6

[63.5,
[73.9,
[57.4,
[56.7,
[85.5,
[31.5,

65.1]
80.0]
63.5]
58.5]
85.8]
35.8]

58.7
75.7
49.9
52.8
95.9
3.6

[57.9, 59.5]
[72.6, 78.8]
[47.1, 52.7]
[51.9, 53.7]
[95.8, 96.0]
[3.3, 3.9]

66.8
74.0
64.9
62.3
88.0
38.8

[65.3,
[67.1,
[59.2,
[60.7,
[87.7,
[35.3,

68.3]
80.9]
70.6]
63.9]
88.3]
42.3]

68.8
76.0
62.9
64.4
96.2
5.8

[67.3, 70.3]
[69.2, 82.8]
[57.2, 68.6]
[62.8, 66.0]
[96.0, 96.4]
[5.2, 6.4]

71.8
76.9
58.7
61.1
83.7
28.5

[69.0,
[69.0,
[49.9,
[57.6,
[82.9,
[20.7,

74.6]
84.8]
67.5]
64.6]
84.5]
36.3]

63.7
83.0
52.6
52.8
95.2
3.9

[60.9, 66.5]
[75.6, 90.4]
[44.2, 61.0]
[49.5, 56.1]
[94.7, 95.7]
[2.7, 5.1]

60.9
77.2
56.6
53.0
83.4
29.5

[59.7,
[73.0,
[52.3,
[51.7,
[83.1,
[26.3,

62.1]
81.4]
60.9]
54.3]
83.7]
32.7]

52.2
72.5
41.7
45.8
95.6
2.5

[51.1, 53.3]
[68.2, 76.8]
[38.0, 45.4]
[44.6, 47.0]
[95.4, 95.8]
[2.2, 2.8]

68.9
86.4
62.9
64.2
89.1
27.5

[65.8,
[76.3,
[50.9,
[60.8,
[88.5,
[18.8,

72.0]
95.5]
74.9]
67.6]
89.7]
36.2]

62.6
80.8
68.4
57.4
97.1
2.7

[59.5, 65.7]
[69.5, 92.1]
[56.3, 80.5]
[54.0, 60.8]
[96.8, 97.4]
[1.7, 3.7]

2007 Blackwell Publishing Ltd. No claim to original US government works. Paediatric and Perinatal Epidemiology, 21 (Suppl. 2), 6271

LMP and ultrasound estimates of gestational age

67

Table 3. Mean birthweight and NICU admissions by cross-tabulation of LMP-baseda and ultrasound-basedb gestational age estimates
Ultrasound
LMP
gestational age (weeks)
2023
n
Mean birthweight (g)
(SD)
% NICU
2427
n
Mean birthweight (g)
(SD)
% NICU
2831
n
Mean birthweight (g)
(SD)
% NICU
3236
n
Mean birthweight (g)
(SD)
% NICU
3741
n
Mean birthweight (g)
(SD)
% NICU
4244
n
Mean birthweight (g)
(SD)
% NICU
45
n
Mean birthweight (g)
(SD)
% NICU
Total
n
Mean birthweight (g)
(SD)
% NICU

2023

2427

2831

3236

3741

4244

45

Total

32
810
(526)
88

83

67

25
3 437
(485)
8

55
590
(255)
96

343
857
(281)
98

59
1291
(430)
97

23
2 164
(531)
56

58
3 409
(516)
2

107
915
(257)
98

616
1395
(364)
97

218
2 211
(524)
63

286
3 347
(435)
3

17
953
(212)
100

272
1595
(391)
93

6568
2584
(522)
40

5 560
3 286
(482)
5

30
3677
(391)
0

12 449
2 876
(643)
25

12
1482
(1220)
80

40
1760
(828)
85

4245
2 873
(495)
18

129 218
3 453
(456)
3

1173
3811
(458)
7

12
3545
(467)
0

134 708
3 437
(471)
4

15
1944
(923)
93

195
2 865
(554)
21

12 539
3 522
(463)
3

538
3828
(485)
6

13 293
3 522
(482)
4

152
2 792
(550)
24

3 225
3 512
(468)
3

76
3831
(571)
7

11 410
2 691
(536)
32

150 911
3 454
(459)
3

1818
3815
(470)
6

17
3441
(547)
0

134
481
(109)
c

100
0
d

0
8
d

38
1

75

100

100

11
1309
(273)
91

211
620
(593)
89

522
887
(367)
97

1019
1463
(433)
95

2
d

0
2
d

206
996
(1035)
66
539
1 213
(900)
84
1 235
1 946
(942)
69

3 478
3 470
(538)
5
165 908

LMP from birth certicate.

Ultrasound from XAFP screening form.
c
Missing data.
d
Birthweight means were not calculated for n < 10.
LMP, last menstrual period; NICU, neonatal intensive care unit.
b

birthweight distributions overlap and appear normally

distributed. For LMP-based gestational age 42 weeks
(gure not shown), the birthweight distribution is
wider, atter, and shifted to the left, compared with the
ultrasound-based distribution.

According to LMP-based gestational age estimates,

8.7% of the infants were preterm (<37 weeks), 81.2%
were term (3741 weeks) and 10.1% were post-term
(42 weeks). In comparison, according to ultrasoundbased estimates, 7.9% of the infants were preterm,

2007 Blackwell Publishing Ltd. No claim to original US government works. Paediatric and Perinatal Epidemiology, 21 (Suppl. 2), 6271

68

P. M. Dietz et al.

Table 4. Proportion of women with

inconsistenta estimates of gestational age
and adjusted odds ratios for inconsistent
estimates by selected maternal and
pregnancy characteristics

Race/ethnicity
White
African American
Asian
Hispanic
Other
Age (years)
<20
2024
2534
35
Education (years)
<12
12
>12
Month prenatal care began
12
34
Parity
0
1
2

Inconsistent (%)

Adjusted ORb [95% CI]

13.2
19.0
13.8
20.5
14.4

Reference
1.3 [1.2, 1.4]
1.1 [1.0, 1.2]
1.3 [1.2, 1.4]
1.1 [1.0, 1.2]

22.5
20.9
16.0
13.5

1.7 [1.6, 1.8]

1.6 [1.5, 1.7]
1.3 [1.2, 1.3]
Reference

22.4
19.0
13.3

1.4 [1.3, 1.4]

1.2 [1.2, 1.3]
Reference

15.3
22.0

Reference
1.5 [1.4, 1.5]

15.5
17.1
19.8

Reference
1.2 [1.1, 1.2]
1.3 [1.3, 1.4]

Inconsistent is >14 days absolute difference between LMP estimate and ultrasound
estimate.
b
Adjusted for all characteristics simultaneously.
a

91.0% were term and 1.1% were post-term. Using

ultrasound as the gold standard, the overall sensitivity
(the percentage of true preterm deliveries correctly
identied by LMP) was 64.3%, and the positive predictive value (the percentage of those found to be preterm
by LMP that were true preterm) was 58.7% (Table 2).
The sensitivity and positive predictive value were
higher for the gestation group of 2027 weeks than the
other preterm groups. They were lowest for the postterm group, with a sensitivity of 33.6% and positive
predictive value of 3.7%. When stratied by race/
ethnicity, Hispanics had the lowest sensitivity and
positive predictive value for less than 37 weeks.
Whereas whites had similar sensitivity and positive
predictive value, Hispanics and African Americans
had lower positive predictive value than sensitivity,
meaning that the number of infants falsely identied as
preterm using LMP estimates exceeded the number of
true preterm infants missed by these estimates.
In order to evaluate ultrasound as a measure of gestational age versus LMP, we compared the mean birthweights of infants with gestational age estimates that
were concordant and discordant, using LMP and ultrasound (Table 3). Among discordant gestational age

groups, mean birthweights categorised by ultrasound

were closer to the mean birthweights of infants with
concordant estimates than those categorised by LMP.
In Table 3, mean birthweights for gestational age categories as determined by ultrasound (columns) were
more similar to one another than were mean birthweights for gestational age categories determined from
the LMP (rows). However, some misclassication
among infants <37 weeks gestation based on ultrasound was apparent (potentially due to clerical error),
as mean birthweights increased with increasing LMPbased gestational age among infants with discordant
estimates. Examination of percentage of infants in the
NICU showed that ultrasound estimates of gestational
age were more consistent with what would be
expected. Preterm gestational age groups determined
by ultrasound had a higher percentage of infants in the
NICU than did those determined by LMP.
Overall, 17.2% of gestational age estimates had an
absolute difference of >14 days between the two
sources (Table 4); for 4.0% of the records, the
ultrasound-based estimate was greater than the LMPbased estimate and for 13.2% the LMP-based estimate
was greater than the ultrasound-based estimate.

2007 Blackwell Publishing Ltd. No claim to original US government works. Paediatric and Perinatal Epidemiology, 21 (Suppl. 2), 6271

LMP and ultrasound estimates of gestational age

LMP
Characteristic

Preterm rate

Race/ethnicity
White
African American
Asian
Hispanic
Other
Age (years)
<20
2024
2534
35
Education (years)
<12
12
>12
Month prenatal care began
12
34
Parity
0
1
2
Infant gender
Female
Male

Ultrasound

UOR [95% CI]

Preterm rate

UOR [95% CI]

7.3
12.5
7.3
9.8
9.2

Reference
1.8 [1.7, 1.9]
1.0 [0.9, 1.1]
1.4 [1.3, 1.4]
1.3 [1.2, 1.4]

7.3
10.8
6.5
8.1
8.4

Reference
1.5 [1.4, 1.6]
0.9 [0.8, 1.0]
1.1 [1.1, 1.2]
1.2 [1.1, 1.2]

10.8
8.7
8.3
10.6

1.0 [0.9, 1.1]

0.8 [0.8, 0.9]
0.8 [0.7, 0.8]
Reference

9.0
7.6
7.6
9.3

1.0 [0.9, 1.0]

0.8 [0.8, 0.9]
0.8 [0.8, 0.8]
Reference

10.4
9.4
7.8

1.4 [1.3, 1.4]

1.2 [1.2, 1.3]
Reference

8.3
8.3
7.5

1.1 [1.1, 1.2]

1.1 [1.1, 1.2]
Reference

8.9
8.6

Reference
1.0 [0.9, 1.0]

8.1
7.3

Reference
0.9 [0.9, 0.9]

8.8
8.0
10.0

1.1 [1.0, 1.1]

Reference
1.3 [1.2, 1.3]

8.3
7.0
8.4

1.2 [1.2, 1.3]

Reference
1.2 [1.2, 1.3]

8.2
9.6

Reference
1.2 [1.1, 1.2]

7.2
8.6

Reference
1.2 [1.2, 1.3]

69

Table 5. Preterm ratesa and UORs using

LMP- and ultrasound-based gestational age
estimates for selected characteristics

Rates are limited to gestational ages between 20 and 44 weeks.

LMP, last menstrual period; UOR, unadjusted odds ratio.

African American and Hispanic women compared

with white women had a greater percentage of records
with inconsistent LMP and ultrasound gestational age,
as did women aged <35 years compared with their
older counterparts, women with fewer years of education compared with women with 13 years, and
multiparae compared with primiparae. Women who,
according to birth certicate records, entered into prenatal care in the third or fourth month of pregnancy
had infants with higher rates of inconsistent estimates
compared with women who entered in the rst or
second month. We found the same groups of women
with higher inconsistent estimates when we stratied
by LMP estimate > ultrasound estimate and LMP
estimate < ultrasound estimate (data not shown).
Preterm delivery rates differed according to maternal characteristics when using LMP and ultrasound
(Table 5). For example, the odds ratio (OR) for preterm
delivery for African American infants compared with
white infants was higher for LMP-based at 1.8 [95% CI
1.7, 1.9] than for ultrasound-based gestational age esti-

mates at 1.5 [95% CI 1.4, 1.6]. A similar pattern was

found for education. The OR of 1.2 for preterm delivery
for male infants compared with female infants was the
same for LMP-based and ultrasound-based gestational
age estimates, and thus there was no evidence that
gestational age based on ultrasound resulted in higher
preterm rates among fetuses known to be smaller, such
as females.

Discussion
Using ultrasound-based gestational age as the gold
standard, this study found evidence of misclassication
of gestational age based on LMP. We found a greater
percentage of false preterm infants, resulting in ination
of the preterm delivery rate. In addition, African Americans and Hispanics had a greater percentage of records
with misclassied gestational age than white women,
resulting in inated racial/ethnic disparities in preterm
rates. The same pattern was found for women with less

2007 Blackwell Publishing Ltd. No claim to original US government works. Paediatric and Perinatal Epidemiology, 21 (Suppl. 2), 6271

70

P. M. Dietz et al.

education. The birthweight distributions for gestational

ages <32 weeks were bimodal when based on LMP but
unimodal when based on ultrasound. While concerns
have been raised that ultrasound-based gestational age
results in misclassication of fetuses smaller than
expected, we found no evidence of this bias in our
study, as the ORs of preterm delivery for male infants
compared with female infants were the same for LMPand ultrasound-based gestational ages. The majority of
inconsistent estimates for LMP-based post-term infants
were of gestational ages greater than those arrived at by
ultrasound. This is consistent with our knowledge that
ovulation is more likely to occur later than the assumed
14 days after the rst day of the LMP rather than earlier.
Our nding that the ultrasound-based gestational
age distribution had fewer post-term deliveries is consistent with those of other studies.9,18 However, ndings regarding preterm rates are not consistent: one
study found higher preterm rates using ultrasound
estimates,18 another found preterm rates to be similar
between ultrasound and LMP estimates,9 while we
found higher preterm rates with LMP-based estimates.
These inconsistent ndings suggest that the amount
and type of error in LMP-based gestational age can
vary depending upon characteristics of the sample and
data collection methods. Some types of error, such as
delayed ovulation, result in overestimation of gestational age, whereas poor recall could cause error in
either direction. The predominant direction of the error
will determine an overall under- or overestimation of
gestational age compared with the true estimate. With
LMP, it is likely that more than one type of error is
affecting the estimate of gestational age and contributing to bidirectional misclassication.
Our study beneted from a large sample size that
included a subpopulation of women from the cohort
who gave birth in California in 2002. While this study
population may be more representative and have more
statistical power than those based on hospital or clinic
samples,9,12,13,18 characteristics of women included in
our sample differed from those not included in several
important ways. Our sample included more women
with post-term gestational age based on LMP, which is
a marker for poor dating. Women screened in the
XAFP programme who received ultrasound were
more likely to have had post-term LMP dates than
those who did not receive ultrasound, suggesting that
uncertain dates might have been an indication for
the ultrasound. Therefore, the LMP-based dates for
women included in our sample may be less reliable

than the LMP-based dates for the general population.

If so, the misclassication rate of gestational age from
LMP could be lower in the general population than
found in this study.
On the other hand, women in our sample were more
educated, less likely to have Medicaid coverage, older
(with the exception of women aged 35 years, who are
eligible for amniocentesis without XAFP screening),
and entered prenatal care earlier on average than
excluded women, attributes associated with more
reliable LMP dates. It is reassuring that LMP-based
preterm rates among included and excluded women
were similar, suggesting that the misclassication of
LMP-based gestational age among preterm infants may
indeed be representative of the general population of
California. Finally, while we assumed ultrasound to be
the gold standard when estimating sensitivity and
positive predictive value, we found some evidence of
error in ultrasound-based gestational age estimates.
Therefore, the sensitivity and positive predictive value
of LMP-based gestational age may be higher for the
entire cohort of infants in California than described in
our sample.
In conclusion, our study provides evidence that a
substantial amount of misclassication results when
using LMP-based gestational age estimates, and this
misclassication can lead to inated preterm delivery
rates. In addition, differences in preterm delivery rates
between whites and African Americans, and between
whites and Hispanics, can also be inated. Including
ultrasound-based estimates of gestational age on the
birth certicate would help to improve the accuracy of
preterm delivery rates, yet not all women receive an
ultrasound before 20 weeks gestation. Those who
receive ultrasound may have uncertain LMP dates (an
indication for ultrasound), and are more likely to be
privately insured. The 2003 revised US standard birth
certicate includes a new gestational age item, the
obstetric estimate, which is the clinicians best estimate
of gestational age at delivery given available dating
information, including ultrasound but excluding neonatal assessments. Validation of this item will be
important to assess whether it helps address the problems presented with LMP-based gestational age.

Acknowledgements
The California Department of Health Services, Genetic
Disease Branch collected the XAFP and Newborn
Screening programme records and the California

2007 Blackwell Publishing Ltd. No claim to original US government works. Paediatric and Perinatal Epidemiology, 21 (Suppl. 2), 6271

LMP and ultrasound estimates of gestational age

Center for Health Statistics provided the birth cohort
les. Allen Hom and Steve Graham of the Sequoia
Foundation conducted the record linkage.

References
1 David RJ. The quality and completeness of birthweight and
gestational age data in computerized birth les. American
Journal of Public Health 1980; 70:964973.
2 Kramer MS, Platt RW, Wen SW, Joseph KS, Allen A,
Abrahamowicz M, et al. A new and improved
population-based Canadian reference for birthweight for
gestational age. Pediatrics 2001; 108:E35.
3 Zhang J, Bowes WA Jr. Birth-weight-for-gestational-age
patterns by race, sex, and parity in the United States
population. Obstetrics and Gynecology 1995; 86:200208.
4 Vahratian A, Buekens P, Bennett TA, Meyer RE, Kogan MD,
Yu SM. Preterm delivery rates in North Carolina: are they
really declining among non-Hispanic African Americans?
American Journal of Epidemiology 2004; 159:5963.
5 Tentoni S, Astol P, De Pasquale A, Zonta LA. Birthweight
by gestational age in preterm babies according to a
Gaussian mixture model. BJOG 2004; 111:3137.
6 Savitz DA, Terry JW Jr, Dole N, Thorp JM Jr, 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.
7 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.
8 Rowland AS, Baird DD, Long S, Wegienka G, Harlow SD,
Alavanja M, et al. Inuence of medical conditions and
lifestyle factors on the menstrual cycle. Epidemiology 2002;
13:668674.

71

9 Wilcox AJ, Dunson D, Baird DD. The timing of the fertile

window in the menstrual cycle: day specic estimates from
a prospective study. British Medical Journal 2000;
321:12591262.
10 Berg AT. Menstrual cycle length and the calculation of
gestational age. American Journal of Epidemiology 1991;
133:585589.
11 ACOG Committee on Practice Bulletins. American College
of Obstetrics and Gynecology Practice Bulletin No. 58.
Ultrasonography in pregnancy. Obstetrics and Gynecology
2004; 104:14491458.
12 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.
13 Morin I, Morin L, Zhang X, Platt RW, Blondel B, Brart G,
et al. Determinants and consequences of discrepancies in
menstrual and ultrasonographic gestational age estimates.
BJOG 2005; 112:145152.
14 Chervenak FA, Skupski DW, Romero R, Myers MK,
Smith-Levitin M, Rosenwaks Z, et al. How accurate is fetal
biometry in the assessment of fetal age? American Journal of
Obstetrics and Gynecology 1998; 178:678687.
15 Johnsen SL, Rasmussen S, Sollien R, Kiserud T. Fetal age
assessment based on ultrasound head biometry and the
effect of maternal and fetal factors. Acta Obstetricia et
Gynecologica Scandinavica 2004; 83:716723.
16 Klln K. Mid-trimester ultrasound prediction of gestational
age: advantages and systematic errors. Ultrasound in
Obstetrics and Gynecology 2002; 20:558563.
17 Larsen T, Nguyen TH, Greisen G, Engholm G, Mller H.
Does a discrepancy between gestational age determined by
biparietal and last menstrual period sometimes signify early
intrauterine growth retardation? BJOG 2000; 107:238244.
18 Yang H, Kramer MS, Platt RW, Blondel B, Brart G, Morin I,
et al. How does early ultrasound scan estimate of gestational
age lead to higher rates of preterm birth? American Journal of
Obstetrics and Gynecology 2002; 186:433437.

2007 Blackwell Publishing Ltd. No claim to original US government works. Paediatric and Perinatal Epidemiology, 21 (Suppl. 2), 6271