Transcript Identifying Pregnancy Deaths in the US: Vital Statistics and

Identifying Pregnancy Deaths in the
United States, 1999-2005
Vital Statistics and Surveillance Systems
Andrea Mackay
National Center for Health Statistics
ESA/STAT/AC.219/30
Background
Deaths from pregnancy complications
Important measure of maternal health
Sentinel public health indicator
A woman’s risk of dying from pregnancy complications has
decreased dramatically over the past century in the U.S.
1999-2005: reported increases in the number and rates of U.S.
maternal deaths, coinciding with:
 Implementation of the Tenth Revision of the International
Classification of Diseases (ICD-10), used to classify the
underlying causes of death
 The 2003 revision of the U.S. Standard Certificate of Death
with a pregnancy status checkbox
The U.S. Standard Certificate of Death (2003 revision)
includes a pregnancy status checkbox
If FEMALE:
 Not pregnant within past year
 Pregnant at time of death
 Not pregnant, but pregnant within 42 days of death
 Not pregnant, but pregnant 43 days to 1 year before death
 Unknown if pregnant within the past year
Not all states have implemented the 2003 revision
In 2005, 18 states had adopted the 2003 revision
Other states used a non-standard checkbox or no checkbox
Non-standard checkbox: no option for deaths within 42 days
Two national sources of data on pregnancy deaths
 National Vital Statistics System (NVSS)





Reports on maternal and late maternal deaths
WHO 42-day post-pregnancy interval for maternal deaths
Deaths coded using ICD-10 and WHO coding guidelines
Uses only information in Part I and Part II of the death certificate
Cause of death: ICD-10 codes O00-95, O98-99, and O96
 Pregnancy Mortality Surveillance System (PMSS)
 Reports on pregnancy-related deaths
 1 year post-pregnancy interval
 Uses all available information on the death certificate and

matched birth certificates and other sources when available
Cause of death: classified into groups, i.e. infection, embolism,
hemorrhage
Deaths per 100,000 live births
Maternal mortality ratio (MMR) and Pregnancy-related
mortality ratio (PMR): United States, 1979-2006
Effects of a pregnancy status check box on the
death certificate
 Improves the identification of maternal deaths using death
certificates
 Makes the death certificate a more valuable source for the
surveillance system
 Reduces the differential between the numbers of pregnancy
deaths identified in CDC’s two data systems
 Can lead to misclassification of some deaths as maternal or
late maternal when the cause of death is not pregnancyrelated
Analysis of U.S. Pregnancy mortality 1999-2005
Combined all pregnancy deaths from NVSS and PMSS to obtain
a combined, unduplicated count:
 Maternal deaths (during or within 42 days of pregnancy)
 Pregnancy-related deaths (during or within 1 year of pregnancy)
Evaluated separately the effects of ICD-10 and the pregnancy
status checkbox:
 Data were stratified into 2 time periods: 1999-2002 and 2003-2005
Some maternal and late maternal deaths in NVSS that matched
to deaths in PMSS were determined to be:
 Not causally related to pregnancy and excluded from the analysis
 Not temporally related to pregnancy and excluded from the analysis
 Misclassified (maternal as late maternal, and vice versa) by
additional temporal information available in PMSS, and were
reclassified appropriately.
Results
Mortality ratios* increased significantly over time
1999-2002
Maternal mortality (MMR)
2003-2005
13.1
15.3
Pregnancy-related mortality (PMR) 14.7
18.1
And were higher than those reported separately by either system
NVSS reported MMR
~9.6
~13.4
PMSS reported PMR
~14.1
~15.8
* Deaths per 100,000 live births
Pregnancy mortality ratios for states categorized by
their death certificate’s pregnancy checkbox status
in 2005: United States, 2002 and 2005
Ratio
Year
Standard
checkbox
Non-standard
checkbox/question
No
pregnancy
checkbox
2002
12.6
12.3
12.2
2005
19.6
14.0
13.8
Absolute change (increase)
7.0
1.7
1.6
Percent change (increase)
55.6%
13.8%
13.1%
2002
15.3
14.0
14.2
2005
22.3
16.7
15.9
Absolute change (increase)
7.0
2.7
1.7
Percent change (increase)
45.8%
19.3%
12.0%
Maternal mortality*
Pregnancy-related mortality*
* Deaths per 100,000 live births
Conclusions
No single system identifies all deaths due to pregnancy.
Combining data from two systems provides a more precise
measurement of maternal mortality.
Mortality ratios increased after the implementation of ICD-10 in 1999:
Between 1995-1997 and 1999-2002
12.9% increase in the MMR
16.6% increase in the PMR
Ratios increased even more with use of the pregnancy checkbox in 2003
Between 1999-2002 and 2003-2005
16.7% in the MMR
23.1% in the PMR
Conclusions
Estimated mortality ratios are higher than:
WHO et al reported U.S. MMR of 11.0 for 2005
Hogan et al (Lancet) estimated U.S. PMR of 17.0 for 2008
Use of a standard format checkbox increases ascertainment of pregnancy deaths.
Mortality ratios increased significantly between 2002 and 2005 in states using
a standard format checkbox in 2005
No significant increase in states without a checkbox in 2005
As more states adopt the 2003 revision of the U.S. Standard Certificate of Death, with
the checkbox, mortality ratios in the U.S. may continue to increase
Significant changes in the leading causes of death
 More maternal deaths from indirect causes of death identified with checkbox
 Cardiovascular complications and non-cardio medical conditions became the
leading causes of pregnancy death
Final analytic data set: deaths due to pregnancy,
1999-2002 and 2003-2005
1999-2002
Total
NVSS only
(not reported in
PMSS)
90
Total
Maternal
PregnancyLate
death
related, but
maternal
(O00-95,
coded outside
Death (O96)
O98-99)
chapter O
89
1
Total
283
Maternal
Late
death
maternal
(O00-95,
death (O96)
O98-99)
206
1404
838
2360
28
n/a
1493
29
Pregnancyrelated, but
coded outside
chapter O
77
n/a
Both PMSS and NVSS 1432
PMSS only
(not coded maternal
or late maternal in
NVSS)
2003-2005
n/a
1377
838
571
838
2231
1244
133
n/a
1450
571
210
571
Identifying Pregnancy Deaths in the
United States, 1999-2005
Vital Statistics and Surveillance Systems
Andrea Mackay
National Center for Health Statistics
Background
Deaths from pregnancy complications
Important measure of maternal health
Sentinel public health indicator
A woman’s risk of dying from pregnancy complications has
decreased dramatically over the past century in the U.S.
1999-2005: reported increases in the number and rates of U.S.
maternal deaths, coinciding with:
 Implementation of the Tenth Revision of the International
Classification of Diseases (ICD-10), used to classify the
underlying causes of death
 The 2003 revision of the U.S. Standard Certificate of Death
with a pregnancy status checkbox
Changes in Coding Under ICD-10
ICD-10 includes a new classification (O96) for late maternal deaths
(those occurring 43-365 days after termination of pregnancy)
ICD-10 coding guidelines allow deaths from indirect causes to be
classified as maternal if the pregnancy is reported in Part I or Part II of
the death certificate
ICD-10 recommends the inclusion of a pregnancy status checkbox on
the death certificate
 Provides a temporal relationship between pregnancy and death
 Can be used in determining a maternal/late maternal death
The U.S. Standard Certificate of Death (2003 revision)
includes a pregnancy status checkbox
If FEMALE:
 Not pregnant within past year
 Pregnant at time of death
 Not pregnant, but pregnant within 42 days of death
 Not pregnant, but pregnant 43 days to 1 year before death
 Unknown if pregnant within the past year
Not all states have implemented the 2003 revision
In 2005, 18 states had adopted the 2003 revision
Other states used a non-standard checkbox or no checkbox
Non-standard checkbox: no option for deaths within 42 days
Two national sources of data on pregnancy deaths
 National Vital Statistics System (NVSS)





Reports on maternal and late maternal deaths
WHO 42-day post-pregnancy interval for maternal deaths
Deaths coded using ICD-10 and WHO coding guidelines
Uses only information in Part I and Part II of the death certificate
Cause of death: ICD-10 codes O00-95, O98-99, and O96
 Pregnancy Mortality Surveillance System (PMSS)
 Reports on pregnancy-related deaths
 1 year post-pregnancy interval
 Uses all available information on the death certificate and

matched birth certificates and other sources when available
Cause of death: classified into groups, i.e. infection, embolism,
hemorrhage
Deaths per 100,000 live births
Maternal mortality ratio (MMR) and Pregnancy-related
mortality ratio (PMR): United States, 1979-2006
Effects of a pregnancy status check box on the
death certificate
 Improves the identification of maternal deaths using death
certificates
 Makes the death certificate a more valuable source for the
surveillance system
 Reduces the differential between the numbers of pregnancy
deaths identified in CDC’s two data systems
 Can lead to misclassification of some deaths as maternal or
late maternal when the cause of death is not pregnancyrelated
EXAMPLE
Information about the temporal relationship to pregnancy
is sometimes missing, with no other indication of a death
related to pregnancy.
Matched Live Birth Certificate from PMSS
For this example, information about the temporal relationship to
pregnancy was available from a matched live birth certificate, which
indicated that the woman delivered, by Cesarean section, 26 days
prior to her death.
A marked pregnancy status checkbox, indicating that death occurred
within 42 days of pregnancy, would have provided a temporal
relationship.
Analysis of U.S. Pregnancy mortality 1999-2005
Combined all pregnancy deaths from NVSS and PMSS to obtain
a combined, unduplicated count:
 Maternal deaths (during or within 42 days of pregnancy)
 Pregnancy-related deaths (during or within 1 year of pregnancy)
Evaluated separately the effects of ICD-10 and the pregnancy
status checkbox:
 Data were stratified into 2 time periods: 1999-2002 and 2003-2005
Some maternal and late maternal deaths in NVSS that matched
to deaths in PMSS were determined to be:
 Not causally related to pregnancy and excluded from the analysis
 Not temporally related to pregnancy and excluded from the analysis
 Misclassified (maternal as late maternal, and vice versa) by
additional temporal information available in PMSS, and were
reclassified appropriately.
Results
Mortality ratios* increased significantly over time
1999-2002
Maternal mortality (MMR)
2003-2005
13.1
15.3
Pregnancy-related mortality (PMR) 14.7
18.1
And were higher than those reported separately by either system
NVSS reported MMR
~9.6
~13.4
PMSS reported PMR
~14.1
~15.8
* Deaths per 100,000 live births
Pregnancy mortality ratios for states categorized by
their death certificate’s pregnancy checkbox status
in 2005: United States, 2002 and 2005
Ratio
Year
Standard
checkbox
Non-standard
checkbox/question
No
pregnancy
checkbox
2002
12.6
12.3
12.2
2005
19.6
14.0
13.8
Absolute change (increase)
7.0
1.7
1.6
Percent change (increase)
55.6%
13.8%
13.1%
2002
15.3
14.0
14.2
2005
22.3
16.7
15.9
Absolute change (increase)
7.0
2.7
1.7
Percent change (increase)
45.8%
19.3%
12.0%
Maternal mortality*
Pregnancy-related mortality*
* Deaths per 100,000 live births
Conclusions
No single system identifies all deaths due to pregnancy.
Combining data from two systems provides a more precise
measurement of maternal mortality.
Mortality ratios increased after the implementation of ICD-10 in 1999:
Between 1995-1997 and 1999-2002
12.9% increase in the MMR
16.6% increase in the PMR
Ratios increased even more with use of the pregnancy checkbox in 2003
Between 1999-2002 and 2003-2005
16.7% in the MMR
23.1% in the PMR
Conclusions
Estimated mortality ratios are higher than:
WHO et al reported U.S. MMR of 11.0 for 2005
Hogan et al (Lancet) estimated U.S. PMR of 17.0 for 2008
Use of a standard format checkbox increases ascertainment of pregnancy deaths.
Mortality ratios increased significantly between 2002 and 2005 in states using
a standard format checkbox in 2005
No significant increase in states without a checkbox in 2005
As more states adopt the 2003 revision of the U.S. Standard Certificate of Death, with
the checkbox, mortality ratios in the U.S. may continue to increase
Significant changes in the leading causes of death
 More maternal deaths from indirect causes of death identified with checkbox
 Cardiovascular complications and non-cardio medical conditions became the
leading causes of pregnancy death
Final analytic data set: deaths due to pregnancy,
1999-2002 and 2003-2005
1999-2002
Total
NVSS only
(not reported in
PMSS)
90
Total
Maternal
PregnancyLate
death
related, but
maternal
(O00-95,
coded outside
Death (O96)
O98-99)
chapter O
89
1
Total
283
Maternal
Late
death
maternal
(O00-95,
death (O96)
O98-99)
206
1404
838
2360
28
n/a
1493
29
Pregnancyrelated, but
coded outside
chapter O
77
n/a
Both PMSS and NVSS 1432
PMSS only
(not coded maternal
or late maternal in
NVSS)
2003-2005
n/a
1377
838
571
838
2231
1244
133
n/a
1450
571
210
571