Journal of Data Science 6(2008), 125-134

Underlying and Multiple Causes of Death in Preterm Infants

Panagiota Kitsantas
George Mason University

Abstract: A limited number of studies have utilized multiple causes of

death to investigate infant mortality patterns. The purpose of the present
study was to examine the risk distribution of underlying and multiple causes
of infant death for congenital anomalies, short gestation/low birth weight
(LBW), respiratory conditions, infections, sudden infant death syndrome
and external causes across four gestational age groups, namely ≤ 23, 24 −
30, 31 − 36, ≥ 37, and determine the extent to which mortality from each
condition is underestimated when only the underlying cause of death is used.
The data were obtained from the North Carolina linked birth/infant death
files (1999 to 2003) and included 4908 death records. The findings of this
study indicate that infants born less than 30 weeks old are more likely (odds
ratio ranging from 1.99 to 6.03) to have multiple causes recorded when the
underlying cause is congenital anomalies, respiratory conditions and infec-
tions in comparison to infants whose gestational age is at least 37 weeks. The
underlying cause of death underestimated mortality for a number of cause-
specific deaths including short gestation/LBW, respiratory conditions, infec-
tions and external causes. This was particularly evident among infants born
preterm. Based on these findings, it is recommended that multiple causes,
whenever available, should be studied in conjunction with the underlying
cause of death data.

Key words: Infant mortality, multiple and underlying causes of death,

preterm infants.

1. Introduction

Infant mortality data have been used to assess disease patterns and long-term
trends, develop policies and design public health prevention and intervention pro-
grams. In most studies, mortality data analysis is conducted using the underlying
cause of death which is defined as “the disease or injury that initiated the train of
events leading directly to death, or the circumstances of the accident or violence,
which produced the fatal injury” (National Center for Health Statistics, 2003)1 .
1
National Center for Health Statistics. (2003). NCHS definitions. Cause-of-Death. Available
at: http://www.cdc.gov/nchs/datawh/nchsdefs/list.htm.
126 Panagiota Kitsantas

Many concerns, however, have been raised over the years about the use of un-
derlying cause of death as means of investigating and evaluating infant mortality.
The use of the underlying-cause-of-death model ignores the role of other condi-
tions and their contribution to death, and its reliability may be adversely affected
as disease evaluation and coding tend to vary among physicians and other health
providers (Nam et al., 1989; Carver et al., 1993). For instance, research has em-
phasized that the coding of disease classification (International Classification of
Diseases, Ninth Revision, 1980) does not adequately measure the impact of short
gestation on infant mortality (Carver et al., 1993; Sowards, 1999), while others
have suggested that the overall disease burden may have been underestimated
due to the exclusive application of the underlying-cause-of-death model (Wall et
al., 2005; Redelings et al., 2006). Consequently, researchers have addressed the
potential use of multiple causes of death (Wall et al., 2005; Redelings et al., 2006)
and recommended that more attention should be given to the mode of identifying
cause of death in studies and especially of those of infant mortality (Nam et al.,
1989).
Overall, in infant mortality studies, the extent to which the distribution of un-
derlying and multiple causes of death differ across various gestational age groups
has not been fully addressed. Given that infants born between 32-36 weeks are
at higher risk for death during infancy (Krameret al., 2000) and their higher fre-
quency of occurrence compared to those born at less than 32 gestational weeks
(Gould et al., 2000), it would be of interest to examine the frequency distribu-
tion and risk for one (underlying cause of death) versus multiple causes of infant
death recordings associated with these gestational age groups and determine how
they compare to infants born very preterm (≤ 30 weeks ) and full term (≥ 37
weeks). Examining such differences may shed some light on the frequency and
under-representation of specific causes of death by gestational age when only the
underlying cause of death is utilized, and provide important information on the
likelihood of multiple cause recordings for certain preterm infants.
The purpose of this exploratory study is (1) to examine the distribution of
underlying and multiple causes of infant death for congenital anomalies, short ges-
tation/low birth weight (LBW), respiratory conditions, infections, sudden infant
death syndrome (SIDS) and external causes across four gestational age groups,
namely ≤ 23, 24-30, 31-36, ≥ 37, and (2) identify those preterm infants that are at
an elevated risk of having multiple cause recordings for a specific underlying cause
of death using linked birth/infant death files from 1999 to 2003. Furthermore,
to show the extent to which mortality from each condition is underestimated by
using the underlying cause of death, frequencies for each of the four causes re-
ported as mentions on the death certificate are compared with listings of these
conditions when recorded as the underlying cause only.
 Causes of Death in Preterm Infants 127

2. Methods

The data were obtained from the North Carolina linked birth/infant death
files from 1999 to 2003. This data set was made available by the North Car-
olina State Center for Health Statistics. The State Center provides a central
collection site for birth records and infant mortality statistics. There were 4908
linked birth/infant death records. Infant deaths (defined as death between 0 and
364 days) were classified into six major categories, including congenital anoma-
lies, short gestation/LBW, respiratory conditions, infections, SIDS and external
causes (i.e., homicide, injuries, unknown causes, conditions originated in the peri-
natal period). They were classified according to the International Classification
of Diseases, tenth revision (ICD-10) (National Center for Health Statistics, 2002).
These causes of death were elected since they constitute some of the leading causes
of infant mortality and they are frequently used in mortality data analyses. Ges-
tational age classifications were as follows: ≤ 23, 24-30, 31-36 and ≥ 37 weeks.
Classification of gestational age was based on findings and recommendations of
previous studies that have examined the high risk of mortality and morbidity
among premature infants (Kramer et al., 2000).
Infant deaths were classified into two categories as having multiple causes of
death (at least one non-underlying cause or mentions) and only one cause (under-
lying cause). Furthermore, in order to determine whether a specific condition has
been underestimated when the underlying cause of death is used, each mention
of the cause was compared to the listing of the condition as the underlying cause
only.
Mortality frequencies and proportions were computed across the six causes
and classifications of gestational age to compare cases with one underlying cause
versus those with multiple causes, and determine the extent to which a specific
cause is being underestimated by the use of underlying cause. Binary logistic
regression was used and 95% confidence intervals are reported to examine the
risk distribution for each gestational age group of having multiple causes versus
one underlying cause. The analyses were performed using SPSS 13.0.

3. Results

Overall, 52.8% of the 4908 infant who died had multiple causes reported in
addition to the underlying cause. The average number of causes per individual
was 3.52 (lower quartile = 2.00, median = 4, upper quartile = 5). Table 1,
which provides information on the frequency distribution and likelihood of cases
with one versus multiple causes of death by gestational age, indicates that overall
infants born between 24 to 30 weeks were approximately 3 times more likely to
die of multiple causes in comparison to full term infants, and they had the highest
128 Panagiota Kitsantas

Table 1: Distribution of cases with only one cause (underlying cause) and multiple
causes of death by gestational age.

Cases with one cause Cases with multiple causes OR 95% CI

n % n %
≤ 23 1131 48.8 613 23.6 0.53 (0.45, 0.61)
24-30 287 12.4 878 33.8 2.99 (2.53, 3.56)
31-36 242 10.5 430 16.6 1.73 (1.43, 2.09)
≥ 37 655 28.3 672 26.0 Reference
Total 2315 2593

Table 2: Underlying and multiple causes of death by gestational age ) and

multiple causes of death by gestational age.

Underlying cause Cases with Cases with OR 95% CI

of death one cause multiple causes
n % n %
Congenital anomalies
≤ 23 4 2.0 53 7.6 4.98 (1.76, 14.1)
24-30 20 9.7 106 15.3 1.99 (1.18, 3.36)
31-36 67 32.5 230 33.1 1.29 (0.91, 1.82)
≥ 37 115 55.8 306 44.0 Reference
Total 206 695
Short gestation/LBW
≤ 23 744 87.5 76 58.0 0.21 (0.02, 2.28)
24-30 100 11.7 54 41.2 1.08 (0.96, 12.2)
31-36 4 0.5 0 0.0 –
≥ 37 2 0.2 1 0.8 Reference
Total 850 131
Respiratory conditions
≤ 23 35 37.0 83 22.6 2.61 (1.02, 6.69)
24-30 46 48.0 252 68.6 6.03 (2.42, 15.0)
31-36 3 3.2 22 6.0 8.07 (1.84, 35.4)
≥ 37 11 11.6 10 2.7 Reference
Total 95 367
 Causes of Death in Preterm Infants 129
Table 2 (continued): Underlying and multiple causes of death by gestational
age ) and multiple causes of death by gestational age.

Underlying cause Cases with Cases with OR 95% CI

of death one cause multiple causes
n % n %
Infections
≤ 23 8 13.6 41 19.0 2.76 (1.02, 7.48)
24-30 22 93.2 118 54.6 2.88 (1.31, 6.38)
31-36 15 25.4 31 14.4 1.11 (0.45, 2.73)
≥ 37 14 23.7 26 12.0 Reference
Total 59 216
SIDS
≤ 23 0 0 0 0
24-30 13 3.0 6 23.1 11.9 (3.90, 36.3)
31-36 85 19.6 7 27.0 2.12 (0.82, 5.48)
≥ 37 335 77.4 13 50.0 Reference
Total 433 26
External causes
≤ 23 340 50.6 360 31.1 0.58 (0.46, 0.75)
24-30 86 12.8 342 29.5 2.27 (1.68, 3.06)
31-36 68 10.1 140 12.1 1.16 (0.82, 1.63)
≥ 37 178 26.5 316 27.3 Reference
Total 672 1158

number of multiple causes recorded on their death certificates. Although infants

with a gestational age of 31-36 weeks experienced the least number of multiple
causes (16.6%), they were still more likely (OR = 1.73) than term infants to have
more than one cause reported. Furthermore, infants born at or less than 23 weeks
were 0.53 times less likely compared to those born term to have additional causes
recorded. This group also had the highest number (48.8%) of one cause (only
underlying cause) recorded on their death certificate.
A further stratification by specific cause of death and gestational age (Table
2) revealed that infants born at 23 weeks or less were more likely (OR ranging
from 2.61 to 4.98) to have multiple causes recorded when the underlying cause
was congenital anomalies, respiratory conditions and infections in comparison
to infants whose gestational age was at least 37 weeks. Similarly, infants with a
gestational age of 24-30 weeks were far more likely (OR ranging from 1.99 to 6.03)
to have multiple causes recorded compared to those born term for congenital
130 Panagiota Kitsantas
Table 3: Underlying cause of death as a percentage of mentions (multiple
causes) by gestational age.

Underlying cause as a
Underlying cause Mentions percentage of mentions
n % n % %
Congenital anomalies
≤ 23 57 6.3 44 8.1 –
24-30 126 14.0 114 20.8 –
31-36 297 33.0 172 31.5 –
≥ 37 421 46.7 216 39.5 –
Total 901 546 –
Short gestation/LBW
≤ 23 820 83.6 217 39.1 –
24-30 154 15.7 283 51.0 54.4
31-36 4 0.4 46 8.3 8.7
≥ 37 3 0.3 9 1.6 33.4
Total 981 555 –
Respiratory conditions
≤ 23 118 25.5 191 23.7 61.7
24-30 298 64.5 448 55.6 66.2
31-36 25 5.4 85 10.6 29.4
≥ 37 21 4.6 81 10.1 26.0
Total 462 805 57.4
Infections
≤ 23 49 17.8 52 27.5 94.2
24-30 140 51.0 47 24.8 –
31-36 46 16.7 42 22.3 –
≥ 37 40 14.5 48 25.4 83.4
Total 275 189 –
External causes
≤ 23 700 38.3 266 15.3 –
24-30 428 23.4 644 37.0 66.5
31-36 208 11.4 328 18.8 63.4
≥ 37 494 30.0 500 28.7 98.8
Total 1830 1738 –
 Causes of Death in Preterm Infants 131

anomalies, respiratory conditions, infections and external causes. For respiratory

related deaths, the odds ratio for cases having multiple causes increased with
increasing gestational age.
The underlying-cause-of-death data captured only 8.7% of short gestation/LBW
related deaths identified by multiple-cause records (mentions) for those born be-
tween 31-36 weeks, and 54.4% for infants with a gestational age of 24-30 weeks
(Table 3). Respiratory related deaths were mostly underestimated by the use of
the underlying cause which captured only 29.4% of cases reported in multiple-
cause data for infants born between 31-36 weeks, and 26.0% for term infants.
Furthermore, the percentage of deaths due to external causes relative to men-
tions was found to be moderate for 24-30 weeks (66.5%) and 31-36 weeks (63.4%).
Deaths due to SIDS were not recorded as mentions for any of the other conditions
or diseases. Congenital anomalies were declared as underlying causes more fre-
quently on these death certificates than mentions. Overall, the underlying cause
of death as a percentage of mentions was the highest for term infants compared
to preterm across most of these major causes of infant mortality.

4. Discussion

Underlying and multiple cause data have been mainly employed to investigate
various causes of death recorded on death certificates (Wall et al., 2005; Redelings
et al., 2006) such as cancer (Richardson, 2006) and tuberculosis (Santo et al.,
2003). These applications clearly indicate that the multiple-cause-of-death model
is superior to just using the underlying-cause-of-death data. Only a handful
number of studies (Nam et al., 1989), however, have examined multiple cause
frameworks of infant deaths. It is essential to indicate empirically any analytical
benefits of using multiple causes of death (Nam at al., 1989), and the extent to
which the underlying cause-of-death data underestimate mortality from specific
diseases or conditions. Furthermore, the assessment of multiple and underlying
causes of death in infants has become even more crucial as concerns have been
raised over the contribution of preterm birth to infant mortality (Kramer et al.,
2000), and the impact of classification systems on the process of finalizing the
underlying cause of death.
In this study, the underlying cause of death underestimated mortality for
a number of cause-specific deaths including short gestation/LBW, respiratory
conditions, infections and external causes. This was particularly evident among
infants born preterm. An under-representation of these causes due to the use
of underlying cause of death may be problematic in the assessment of infant
mortality or applications that utilize infant mortality data to study premature
births and LBW (especially for those born between 24-30 weeks) and respiratory
related problems. It also can affect prevention programs designed to prevent
132 Panagiota Kitsantas

deaths associated with such causes in premature infants.

Furthermore, whenever a respiratory condition was the underlying cause of
death, the odds of frequent recording of multiple causes increased with gestational
age. It is known that infants born at lower gestational ages have an increased
frequency and severity of respiratory illness (Weisman, 2003). However, little
is known about the contribution of respiratory illnesses associated with other
conditions to mortality for specific gestational age groups such as those born at
31 weeks and above. Further research using multiple causes of death may warrant
the relationship between gestational age and respiratory related deaths.
Overall, regardless of the specific cause of death, infants born at or less than
23 weeks of gestation were less likely to have multiple causes reported on their
death certificates in comparison to full term infants. Further stratification by
specific cause of death, however, revealed that they were more likely to experi-
ence multiple conditions compared to their term counterparts. This difference
between the cumulative underlying-cause-of-death model (all causes of death)
and the stratified one may be explained by the large number of recorded deaths
due to external and short gestation/LBW related causes which masked the likeli-
hood of recorded multiple conditions for this specific gestational age group. This
indicates that researchers should explicitly select cause-specific-of-death models
when investigating infant mortality since the recordings of other contributing
causes to death may vary by the underlying cause of death.
Previous research has shown that the vast majority of deaths are attributable
to complications of short gestation developed during the neonatal period (Swarz,
1990; Kliegman et al., 1990). This may explain the large number of multiple
causes recorded for preterm infants as they are more prone to develop other
conditions and diseases which are likely to lead to death compared to term infants.
This in turn complicates the analyses of causes of death for premature infants
as the underlying cause of death may oversimplify the biological mechanisms of
death for these infants. Consequently, researchers should carefully contemplate
the analysis of multiple causes for premature infants since the underlying cause
of death alone may not adequately reflect the contribution of other important
conditions to mortality and morbidity.
There are some limitations, however, in using death certificate data. Death
certificates are often imprecise and data are not rigorously evaluated (Nielsen et
al., 1991). Variation also exists in opinions among individuals who are responsi-
ble for “coding” a condition or a disease and that can impact the accuracy of the
classification system (Nam et al., 1989; Carver et al., 1993). Nonetheless, death
certificate data are important in studying disease trends and patterns and can
be used to prevent death. The few research studies that have been conducted
on multiple and underlying causes of death suggest that multiple-causes-of-death
 Causes of Death in Preterm Infants 133

analyses can enrich and help define the mode of death more explicitly than the
underlying-cause-of-death model. As Nam et al. (1989) indicate it is important
for epidemiologists and public health officials to recognize and take into consid-
eration that infant death can be the result of a number of causes rather than one
cause. Based on the findings of this study, it is recommended that multiple causes
are studied in conjunction with the underlying cause among premature infants
and especially for those born between 32 and 36 weeks of gestational age. This
can lead to better understanding of morbidity and mortality in infants and can
assist in the development of more effective prevention and intervention programs
targeting infant mortality.

References

Carver, J. D., McDermott, R. J., Jacobson, H. N., Sherin, K. M., Kanrek, K., Pimentel,
B. and Tan, L. H. (1993). Infant mortality statistics do not adequately reflect the
impact of short gestation. Pediatrics 92, 229-232.
Gould, J. B., Benitz, W. E. and Liu, H. (2000). Mortality and time to death in very
low birth weight infants: California, 1987 and 1993. Pediatrics 105, 1-5.
World Health Organization (1980). International Classification of Diseases, Ninth Re-
vision. World Health Organization, Geneva.
Kliegman, R., Rottman, C. J. and Behrman, R. E. (1990). Strategies for the prevention
of low birth weight. American Journal of Obstetrics and Gynecology 162, 1073-
1083.
Kramer, M. S., Demissie, K., Yang, H., Platt, R. W., Sauve, R. and Liston, R. (2000).
The contribution of mild and moderate preterm birth to infant mortality. The
Journal of American Medical Association 284, 843-849.
Nam, C. B., Eberstein, I. W., Deeb, L. C. and Terrie, E. W. (1989). Infant mortality
by cuase: A comparison of underlying and multiple cause designation. European
Journal of Population 5, 45-70.
National Center for Health Statistics (1997-2004). Data File Documentations, Multiple
Cause-of-Death, 2000-2001. National Center for Health Statistics.
Nielsen, G. P., Bjornsson, J. and Jonasson, J. G. (1991). The accuracy of death certifi-
cates. Virchows Archives [A] 419, 143-146.
Redelings, M. D., Sorvillo, F. and Simon, P. (2006). A comparison of underlying cause
and multiple causes of death: US vital statistics, 2000-2001. Epidemiology 17,
100-103.
Richardson, D .B. (2006). Use of multiple cause of death data in cancer mortality
analyses. American Journal of Industrial Medicine 49, 683-689.
134 Panagiota Kitsantas

Santo, A. H., Pinheiro, C. E. and Jordani, M. S. (2003). Multiple-causes-of-death

related to tuberculosis in the state of Sao Paulo, Brazil, 1998. Revista de Saude
Publica 37, 714-721.
Sowards, K. A. (1999). What is the leading cause of Infant mortality? A note on the
interpretation of official statistics. American Journal of Public Health 89, 1752-
1754.
Swartz, M. (1990). Infant mortality: agenda for the 1990s. Journal of Pediatric Health
Care 4, 169-174.
Wall, M. M., Huang, J., Oswald, J. and McCullen, D. (2005). Factors associated with
reporting multiple causes of death. BMC Medical Research Methodology 5, 1-13.
Weisman, L. E. (2003). Population at risk for developing respiratory syncytial virus
and risk factors for respiratory syncytial virus severity: infants with predisposing
conditions. Pediatric Infectious Disease Journal 22, S33-S39.
Received October 9, 2006; accepted December 9, 2006.

Panagiota Kitsantas
George Mason University
The College of Health and Human Services
Department of Health Administration and Policy, MS 1J3
4400 University Drive
Fairfax, VA 22030, USA
pkitsant@gmu.edu