American Journal of Public Health | March 2005, Vol 95, No. 3478 | Research and Practice | Peer Reviewed | Horon
RESEARCH AND PRACTICE
Objectives. I studied the extent to which maternal deaths are underreported on
Methods. We collected data on maternal deaths from death certificates, link-
age of death certificates with birth and fetal death records, and review of med-
ical examiner records.
Results. Thirty-eight percent of maternal deaths were unreported on death cer-
tificates. Half or more deaths were unreported for women who were undelivered
at the time of death, experienced a fetal death or therapeutic abortion, died more
than a week after delivery, or died as a result of a cardiovascular disorder.
Conclusions. The number of maternal deaths is substantially underestimated
when death certificates alone are used to identify deaths, and it is unlikely that
the Healthy People 2010 objective of reducing the maternal mortality rate to no
more than 3.3 deaths per 100000 live births by 2010 can be achieved. Increasing
numbers of births to older women and multiple-gestation pregnancies are likely
to complicate efforts to reduce maternal mortality. (Am J Public Health. 2005;95:
Underreporting of Maternal Deaths on Death Certificates
and the Magnitude of the Problem of Maternal Mortality
| Isabelle L. Horon, DrPH
examiner records to identify unreported
maternal deaths. The overall undercount of
maternal deaths was estimated, sources of
identification of deaths were reported, and
the degree of underreporting among sub-
groups of the population was described.
The World Health Organization (WHO)
definition of a maternal death was used to
identify deaths for inclusion in this study be-
cause this is the definition used by the Na-
tional Center for Health Statistics (NCHS) to
compile national maternal mortality statistics.
WHO defines a maternal death as “the death
of a woman while pregnant or within 42 days
of termination of pregnancy, irrespective of
the duration and the site of the pregnancy,
from any cause related to or aggravated by
the pregnancy or its management but not
from accidental or incidental causes.”1 1This
definition includes deaths assigned to the
cause “complication of pregnancy, childbirth
and the puerperium” (International Classifica-
tion of Diseases, Ninth Revision [ICD-9] codes
630–6761 1and Tenth Revision [ICD-10]
codes O00–O95, O98–O99, and A34).12
The magnitude of the problem of maternal
mortality is underestimated when mortality
rates are based only on maternal deaths re-
ported on death certificates. Studies have shown
that physicians completing death certificates
after a maternal death fail to report that the
woman was pregnant or had a recent preg-
nancy in 50% or more of these cases.1–3Be-
cause a history of pregnancy must be recorded
on a death certificate for a death to be coded as
resulting from a maternal cause, these deaths
are not included in the calculation of maternal
mortality rates. This leads to an underestimation
of the problem of maternal mortality on both
the state and national levels because death cer-
tificate data collected by states are used to com-
pute maternal mortality rates for the nation.
Previous research has shown that the com-
pleteness of reporting of deaths related to
pregnancy can be improved by linking death
records of women of reproductive ages with
birth and fetal death records2,4–6and through
the use of a check box on the death certifi-
cate to indicate that a decedent was pregnant
at the time of death or had recently been
pregnant.7Other studies have shown that re-
view of medical examiner records is success-
ful in identifying deaths that were not ascer-
tained through other sources, particularly
those among women who were pregnant at
the time of death.8,9Studies that have used
medical examiner records to study preg-
nancy mortality have focused on pregnancy-
associated deaths, defined as deaths from any
cause during pregnancy or within 1 calendar
year of delivery or pregnancy termination.10
Because maternal deaths, as defined below,
are a subset of pregnancy-associated deaths,
review of medical examiner records to iden-
tify maternal deaths should also improve the
completeness of maternal death reporting.
The purpose of this study was to determine
the extent to which maternal deaths are un-
derreported on death certificates by using
both linkage of records and review of medical
Data were collected from 3 sources:
(1) review of death certificates to identify
those records on which a complication of
pregnancy, childbirth, or the puerperium was
listed as an underlying or contributing cause
of death; (2) linkage of death certificates of
reproductive-age women with live birth and
fetal death records to identify a delivery
within 42 days of death; and (3) review of
medical examiner records for evidence that a
woman was pregnant at the time of death or
experienced a recent pregnancy. Data were
collected for all maternal deaths occurring
during the years 1993 through 2000.
Vital records data were obtained from the
Vital Statistics Administration of the Maryland
Department of Health and Mental Hygiene.
Death records were identified by searching for
records on which a complication of pregnancy,
childbirth, or the puerperium was listed as an
underlying or contributing cause of death. For
the years 1993 through 1998, this included all
deaths with ICD-9 codes 630 through 676.
Because Maryland and the remainder of states
began using ICD-10 codes beginning with
1999 mortality data,13all deaths with ICD-10
codes O00 to O95, O98 to O99, or A34 were
included for the years 1999 and 2000.
March 2005, Vol 95, No. 3 | American Journal of Public Health Horon | Peer Reviewed | Research and Practice | 479
RESEARCH AND PRACTICE
Note.The gray area highlights the number of maternal deaths that were identified through death records.The nonshaded
areas show the cases that were missed.
FIGURE 1—Number of maternal deaths, by source of identification, Maryland, 1993–2000.
Changes were made in the classification of
maternal deaths between ICD-9 and ICD-10.
ICD-9 classified a death as having a maternal
cause only if pregnancy was reported as part
of the sequence of events leading to death.
These deaths are classified as maternal in
ICD-10 as well. However, the coding rules for
ICD-10, unlike the rules for ICD-9, classify
deaths aggravated by pregnancy as maternal
deaths. This includes deaths from previously
existing diseases and deaths from nonobstetric
conditions that developed during pregnancy
and were aggravated by physiological effects
of pregnancy.13To account for this discontinu-
ity in the rules for classifying deaths as having
a maternal cause, records for all deaths occur-
ring before 1999 were recoded using ICD-10
rules. The study group therefore includes 16
deaths occurring between 1993 and 1998
that would not have been classified as mater-
nal deaths using ICD-9 rules. Late maternal
deaths (deaths occurring 43 days through
1 year after termination of pregnancy) were
not included in the study group.
Identification of maternal deaths through
linkage of vital records was performed by
matching death certificates for all women of
reproductive age against live birth and fetal
death records to identify pregnancies occur-
ring within 42 days of death. Records were
linked by matching either the mother’s social
security number or the mother’s name and
date of birth on the death record with corre-
sponding information on live birth and fetal
death records. All linked records were manu-
ally reviewed to ensure accurate matching.
Medical examiner records were reviewed
for all women aged 10 through 50 years who
died between 1993 and 2000. Death certifi-
cates were obtained for all women for whom
medical examiner records identified an unde-
livered or recent pregnancy.
All death certificates that were identified
through linkage of records or review of
medical examiner records were reviewed
by a team of 3 board-certified obstetrician-
gynecologists and 2 trained nosologists to de-
termine the underlying cause of death that
would have been assigned if a history of preg-
nancy had been reported on the death certifi-
cate. All records meeting the WHO definition
of a maternal death were included in the
The distribution of maternal deaths identi-
fied through death records alone, and deaths
identified from all sources, were compared by
outcome of pregnancy, time of death, cause
of death, maternal race, age, education and
marital status, parity, and plurality.
A total of 129 maternal deaths occurring
between 1993 and 2000 were identified
from the 3 data sources. Only 80 of these
deaths (62.0%) were identified through
cause-of-death information obtained from
death records (Figure 1). Inclusion of all 129
identified maternal deaths resulted in a ma-
ternal mortality rate of 22.2 per 100000 live
births for the years 1993 through 2000, sub-
stantially higher than the rate of 13.8 per
100000 live births based on information re-
ported on death records alone.
A similar number of maternal deaths were
identified from each of the 3 data sources
(Figure 1). Eighty deaths (62.0%) were identi-
fied through death records, 80 (62.0%)
through linkage of records, and 83 (64.3%)
through review of medical examiner records
(Figure 1). Although most deaths were identi-
fied through more than 1 source, 41 deaths
(31.8%) were identified through only a single
data source. These included 10 deaths (7.8%)
identified through death records, 13 deaths
(10.1%) identified through linkage of records,
and 18 deaths (14.0%) identified through re-
view of medical examiner records.
The number of maternal deaths following
a live birth increased from 50 to 80 and the
number of deaths following a fetal death in-
creased from 4 to 8 when multiple data
sources were used to identify maternal
deaths. Death records identified all 8 deaths
that occurred as a result of an ectopic preg-
nancy and the single death that occurred as a
result of a molar pregnancy but neither of the
2 deaths that followed a therapeutic abortion.
Death records identified only 10 of the re-
maining 23 deaths among women who were
pregnant at the time of death (Table 1).
Among women who were no longer preg-
nant at the time of death, the percentage of
unreported deaths increased with the length of
time between delivery or pregnancy termina-
tion and death. Six of 28 deaths (21.4%) oc-
curring within 1 day of delivery or pregnancy
termination and 6 of 24 deaths (25.0%) oc-
curring 2 through 7 days after delivery were
American Journal of Public Health | March 2005, Vol 95, No. 3 480 | Research and Practice | Peer Reviewed | Horon
RESEARCH AND PRACTICE
TABLE 1—Number of Maternal Deaths, by Source of Identification, Outcome of Pregnancy,
Time of Death, and Cause of Death: Maryland, 1993–2000
Source of Identification
All Maternal Deaths Unreported
on Death Records (%) Sourcesa
Outcome of pregnancy
All other undelivered
Days from delivery to death
Cause of death
aIncludes death records,linkage of records,and medical examiner records.
bIncludes 5 deaths resulting from ectopic pregnancies,1 death resulting from a molar pregnancy,and 22 deaths among
other women who were undelivered at the time of death.
cIncludes cardiomyopathy,congenital heart disease,pulmonary hypertension,endocarditis,valvular dysfunction,and other
cardiac conditions related to or aggravated by pregnancy.
not reported on death records. The percentage
of unreported maternal deaths rose to 75.0%
for deaths occurring 31 to 42 days after deliv-
ery or termination. Twelve of the 28 deaths
(42.8%) that occurred among women who
were pregnant at the time of death were not
reported on death records. This figure in-
cluded 5 deaths resulting from ectopic preg-
nancies, 1 death resulting from a molar preg-
nancy, and 22 deaths among other women
who were undelivered at the time of death.
Deaths were underreported on death rec-
ords for all leading causes of maternal death.
The percentage of unreported deaths was
highest for cardiovascular disorders (56.4%),
followed by embolism and infection (37.5%
each), hypertensive disorders of pregnancy
(21.1%), and hemorrhage (11.8%).
On the basis of information reported on
death certificates, it appeared that cardiovas-
cular disorders, embolism, hemorrhage, and
hypertensive disorders of pregnancy were
each responsible for a similar proportion of
maternal deaths. However, when previously
unreported deaths were included, cardiovas-
cular disorders were clearly the leading cause
of maternal death, responsible for 39 of 129
deaths (30.2%). Embolism, the second lead-
ing cause of death, was responsible for 18.6%
of deaths, whereas hypertensive disorders of
pregnancy, the third leading cause, were re-
sponsible for 14.7% of deaths.
Maternal deaths were underreported for all
categories of maternal race, age, education,
marital status, parity, and plurality (Table 2).
The percentage of unreported deaths was
particularly high for women at the extremes
of the maternal age distribution; half of all
maternal deaths among teenagers and more
than half of all maternal deaths among
women aged 40 and above were unreported.
Today, maternal deaths are relatively rare
events in developed countries. Nevertheless,
maternal deaths still occur, frequently among
young, apparently healthy women, and they
have a devastating impact on the families left
behind. Furthermore, for each woman who
dies, many more experience life-threatening
and often long-lasting complications.1 4
This study supports the findings of earlier
studies that have shown that the number of
maternal deaths is substantially underesti-
mated when death certificates alone are used
to identify deaths. In Maryland, collection of
maternal death data from multiple sources
showed that the maternal mortality rate in
Maryland for the years 1993 through 2000
was 22.2 per 100000 live births, 60.9%
higher than the rate of 13.8 per 100000
based only on information reported on death
records. If maternal deaths are assumed to be
underreported at the same level nationally as
they are in Maryland, the maternal mortality
rate for the United States for the year 2001
would have been 15.9 per 100000 live births,
substantially higher than the reported figure of
9.9 per 100000.15Because it is possible that
not all maternal deaths were identified in this
study even by using additional data sources,
the adjusted rates of 15.9 per 100000 for the
nation and 22.2 per 100000 for Maryland
may also be underestimates of the true figures.
It is unlikely that the Healthy People 2010
objective of reducing the maternal mortality
rate to no more than 3.3 deaths per 100000
live births by 201016can be achieved, espe-
cially because the number of women in 2
groups at increased risk of maternal death—
women of advanced maternal age and
women with multiple-gestation pregnancies—
March 2005, Vol 95, No. 3 | American Journal of Public HealthHoron | Peer Reviewed | Research and Practice | 481
RESEARCH AND PRACTICE
TABLE 2—Number of Maternal Deaths, by Source of Identification and Selected Maternal
and Pregnancy Characteristics: Maryland, 1993–2000
Source of Identification
Death Records (%)
All other races
Maternal education (aged ≥20 y)b
College graduate/graduate school
Live birth order
4th child and over
Twin or multiple
aIncludes deaths identified from death records,record linkage,and review of medical examiner records.
bFigures include data for women aged 20 years and older because younger women may not have completed their education
at the time of death.
cNumber of fetuses in a pregnancy.
has been increasing. Between 1990 and
2001, the US birth rate increased by 47% for
women aged 40 to 44 and tripled for women
aged 45 to 49, whereas the twin birth rate
increased by 33% and the rate of triplet and
higher order births rose by nearly 300%.1 7
Both the increase in births among older
women and the increase in multiple-gestation
pregnancies are attributable in large part to
the increased use of fertility-enhancing thera-
pies. Data compiled in the current study
showed that the maternal mortality rate for
women aged 40 years and older was 84.5
per 100000 live births, more than 4 times
higher than the rate of 20.7 per 100000 for
all younger women. The maternal mortality
rate for women experiencing multiple-gestation
pregnancies was 38.5 per 100000, more
than double the rate of 15.5 per 100000 for
women with known singleton pregnancies.
However, the true gap between maternal
mortality rates for women with singleton and
multiple-gestation pregnancies is not as large
as these figures would suggest because most
of the 35 pregnancies of unknown plurality
were likely to have been singleton pregnan-
cies. Nevertheless, even if all pregnancies of
unknown plurality are assumed to have been
singleton pregnancies, the recalculated mater-
nal mortality rate of 21.4 per 100000 for
women with singleton pregnancies would re-
main substantially lower than the rate for
women with multiple-gestation pregnancies.
Although several studies based on interna-
tional data have also shown that multiple ges-
tation increases maternal mortality,18–20this
association has not previously been shown
using US data. Additional study using US data
is needed to further explore the association
between multiple gestation and maternal
mortality because the increasing number of
multiple births is likely to complicate efforts
to reduce maternal mortality.
The findings of this report show that car-
diovascular disorders, which include condi-
tions such as cardiomyopathy, congenital
heart disease, pulmonary hypertension, endo-
carditis, valvular dysfunction, and other car-
diac conditions related to or aggravated by
pregnancy, are the leading cause of maternal
death in Maryland. This is in contrast to na-
tional death data compiled by the NCHS,
which show the leading causes of maternal
death to be hypertensive disorders of preg-
nancy,hemorrhage, and embolism.15The
Centers for Disease Control and Prevention’s
Pregnancy-Related Mortality Surveillance Sys-
tem (PMSS), which compiles national data on
pregnancy-related deaths, has historically
identified the same 3 leading causes of
death.21–23The PMSS data on pregnancy-
related deaths, which are defined as all
deaths causally related to pregnancy, are
based largely on death certificate data pro-
vided by state vital records offices. It is likely
that cardiovascular disorders have not been
identified as a leading cause of maternal
death in either NCHS or PMSS data because
death records of women dying as a result of
this cause frequently do not indicate that they
were pregnant or had recently been pregnant.
Fewer than half of all deaths resulting from
cardiovascular disorders were identified from
death records in the present study.
It is critical that physicians who care for
pregnant woman are aware that a pregnant
patient or a patient who has recently given
American Journal of Public Health | March 2005, Vol 95, No. 3 482 | Research and Practice | Peer Reviewed | Horon
RESEARCH AND PRACTICE
birth is more likely to die as a result of a car-
diovascular disorder than from any other
cause. Cardiovascular disorders may be of
particular concern for adolescents; this cause
was responsible for 6 of the 10 deaths among
14- to 19-year-olds in this study.
This study has also shown that a larger pro-
portion of maternal deaths occur among un-
delivered women than previously reported.
Although deaths resulting from an ectopic or
molar pregnancy were well-reported on Mary-
land death certificates, more than half of the
deaths that occurred among other undelivered
women were unreported. Deaths among this
subgroup of undelivered women represented
19.3% of all maternal deaths for which the
time of death was known, compared with a
figure of 11.7% in a recent PMSS report.24
The lack of complete reporting of maternal
deaths has led to misconceptions regarding
the magnitude of the problem of maternal
deaths, the leading cause of death, and the
timing of maternal deaths. Death records are
an important source of data on pregnancy
mortality, but death records alone identify
only a fraction of all maternal deaths. New
York City and 17 states have attempted to im-
prove ascertainment of pregnancy on death
records by including a pregnancy check box
or asking about pregnancy status on their
death records. In Maryland, questions about
pregnancy status in the 12 months preceding
death, the outcome of pregnancy, and the
date of delivery were added to the Certificate
of Death in 2001. The NCHS has recom-
mended use of a single pregnancy question
by all states on the revised US Standard Cer-
tificate of Death, but it is likely to be a num-
ber of years before all states begin using the
revised certificate. Currently, comprehensive
identification of maternal deaths can be ac-
complished only by collecting information
from multiple data sources. Both data linkage
and review of medical examiner records con-
tributed substantially to identification of ma-
ternal deaths in Maryland. Linkage of records
identified 13 deaths that were not identified
through death records or review of medical
examiner records. Review of medical exam-
iner records identified 18 deaths that could
not be identified through death records or
linkage of records, including 56% of all
deaths among women who were undelivered
at the time of death, 21% of embolism
deaths, 18% of cardiovascular deaths, and
both deaths that followed therapeutic abor-
tions. Review of paper copies of medical ex-
aminer records to identify maternal deaths
can be a labor-intensive process. Fortunately,
medical examiner records are increasingly be-
coming computerized, which will make the
identification of women who were pregnant
at the time of death or were recently preg-
nant a far less time-consuming process. We
hope that this will encourage the use of med-
ical examiner records for routine surveillance
of deaths related to pregnancy.
Comprehensive identification of maternal
deaths is necessary to determine the magni-
tude of maternal mortality, identify the major
causes of death, and identify groups at in-
creased risk of death. Without a clear under-
standing of these factors, it is not possible to
develop comprehensive strategies to prevent
this devastating pregnancy outcome.
About the Author
The author is with the Vital Statistics Administration at
the Maryland Department of Health and Mental Hygiene,
Requests for reprints should be sent to Isabelle L. Horon,
DrPH, Vital Statistics Administration, Maryland Depart-
ment of Health and Mental Hygiene, 201 West Preston
Street, Baltimore, MD 21201 (email: firstname.lastname@example.org.
This article was accepted May 18, 2004.
The author gratefully acknowledges Dr Diana Cheng
for her guidance, support, and thoughtful review of this
article, and Dr Robert Hayman for assistance with link-
age of data.
Human Participant Protection
No protocol approval was needed for this study.
1. Dye TD, Gordon H, Held B, Tolliver NJ, Holmes AP.
Retrospective maternal mortality case ascertainment in
West Virginia, 1985 to 1989. Am J Obstet Gynecol.
Pregnancy-related mortality: Georgia, 1990–1992.
MMWR Morb Mortal Wkly Rep. 1995;44:93–96.
Centers for Disease Control and Prevention.
tality in developed countries: not just a concern of the
past. Obstet Gynecol. 1995;86:700–705.
Atrash HK, Alexander S, Berg CJ. Maternal mor-
of maternal deaths: Washington State. MMWR Morb
Mortal Wkly Rep. 1986;35:621–623.
Starzyk P, Frost F, Kobayashi JM. Misclassification
Post-delivery mortality in Tennessee, 1989–91. Obstet
Jocums SB, Berg CJ, Entman SS, Mitchell EF.
related mortality: Georgia, 1990–92. MMWR Morb
Mortal Wkly Rep. 1995;44:93–97.
Floyd V, Hadley C, Lavoie M, Toomey K. Pregnancy-
mortality surveillance: Puerto Rico, 1989. MMWR
Morb Mortal Wkly Rep. 1991;40:521–523.
Comas A, Navarro A, Carrera A, et al. Maternal
of maternal deaths in New York City. Am J Public
Allen MH, Chavkin W, Jarinoff J. Ascertainment
pregnancy-associated mortality: Maryland,
1993–1998. JAMA. 2001;285:1455–1459.
Horon IL, Cheng D. Enhanced surveillance for
10. Atrash HK, Rowley D, Hogue CJR. Maternal and
perinatal mortality. Curr Opin Obstet Gynecol. 1992;4:
of Diseases, Injuries, and Causes of Death. Geneva, Switz-
erland: World Health Organization; 1977.
Manual of the International Statistical Classification
12. International Statistical Classification of Diseases
and Related Health Problems, Tenth Revision (ICD10).
Geneva, Switzerland: World Health Organization;
13. Hoyert DL, Arias E, Smith BL, Murphy SL,
Kochanek KD. Deaths: final data for 1999. Natl Vital
Stat Rep. September 21, 2001;49(8).
14. Hebert PR, Reed G, Entman SS, Mitchel EF Jr,
Berg C, Griffin MR. Serious maternal morbidity after
childbirth: prolonged hospital stays and readmissions.
Obstet Gynecol. 1999;94:942–947.
15. Arias E, Anderson RN, Hsiang-Ching K, Murphy SL,
Kochanek KD. Deaths: final data for 2001. Natl Vital
Stat Rep. September 18, 2003:52.
16. Healthy People 2010 Volume II: Objectives for Im-
proving Health. Washington, DC: US Government Print-
ing Office; 2000.
Park MM, Sutton PD. Births: final data for 2001. Natl
Vital Stat Rep. December 18, 2002:51.
Martin JA, Hamilton BE, Ventura SJ, Menaker F,
18. McDermott JM, Steketee R, Wirima J. Mortality
associated with multiple gestation in Malawi. Int J Epi-
19. Blickstein I. Maternal mortality in twin gestations.
J Reprod Med. 1997;42:680–684.
20. Conde-Agudelo A, Belizan JM, Linkmark G. Ma-
ternal morbidity and mortality associated with multiple
gestations. Obstet Gynecol. 2000;95:899–904.
21. Atrash HK, Koonin LA, Lawson HW, Franks AL,
Smith JC. Maternal mortality in the United States,
1970–1986. Obstet Gynecol. 1990;76:1055–1060.
22. Berg CJ, Atrash HK, Koonin LM, Tucker M.
Pregnancy-related mortality in the United States,
1987–1990. Obstet Gynecol. 1996;88:161–167.
23. Chang J, Elam-Evans LD, Berg CJ, et al. Preg-
nancy-related mortality surveillance: United States,
1991–1999. MMWR Surveill Summ. 2003;52:1–8.
24. Berg CJ, Chang J, Callaghan WM, Whitehead SJ.
Pregnancy-related mortality in the United States,
1991–1997. Obstet Gynecol. 2003;101:289–296.