IJE Advance Access originally published online on March 23, 2006
International Journal of Epidemiology 2006 35(4):1083-1089; doi:10.1093/ije/dyl047
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Post-neonatal mortality in Norway 1969–95: a cause-specific analysis
1 Faculty of Social Science, Vestfold University College, PO Box 2243, N-3103 Tønsberg, Norway
2 Department of Mathematics, University of Oslo, PO Box 1053 Blindern, N-0316 Oslo, Norway
3 Section for Epidemiology and Medical Statistics, Department of Public Health and Primary Health Care, University of Bergen, Kalfarveien 31, 5018 Bergen, Norway
4 Division of Epidemiology, Norwegian Institute of Public Health, PO Box 4404 Nydalen, N-0403 Oslo, Norway
* Corresponding author. E-mail: Annett.Arntzen{at}hive.no
 |
Abstract |
|---|
 Top Abstract Materials and methodsResultsDiscussionConclusionsReferences |
|---|
Background We recently reported increased social inequality for post-neonatal death. The aim of the present study was to investigate the association between socioeconomic status and cause-specific post-neonatal death.
Methods All 1 483 857 live births recorded in the Medical Birth Registry of Norway from 1969–95 with information on parents' education were included. During the post-neonatal period (from 28 to 364 days of life) 4464 infants died. Differences between education groups were estimated as risk differences, relative risks, population attributable fractions, and relative index of inequality.
Results The major causes of death were congenital conditions, sudden infant death syndrome (SIDS), and infections. Post-neonatal mortality declined from 3.2/1000 in the 1970s to 1.9/1000 in the 1990s, mainly due to reduced mortality from congenital conditions. The absolute risk for SIDS increased by 0.51/1000 in the same period among infants whose mothers had low education, while it decreased by 0.56/1000 for those whose mothers had high education. The relative risk for SIDS among infants whose mothers had low education increased from 1.02 in the 1970s to 2.39 in the 1980s and 5.63 in the 1990s. Among infants whose fathers were not recorded in the Birth Registry, the absolute risk of SIDS increased by 0.79/1000 from the 1970s to the 1990s.
Conclusions Increased social inequality for post-neonatal death was primarily due to increases in the absolute and relative risks of SIDS among infants whose mothers have low education. Social inequality widened during the study period for SIDS and deaths caused by infections.
Keywords Post-neonatal mortality, causes of death, socioeconomic factors, trend
Accepted 2 March 2006
The risk of infant death has decreased considerably in all socioeconomic groups during the last decades in Norway.1 Nevertheless, the inverse association between parents' education and risk of infant death persists.2 A recent study showed a noticeable decrease in socioeconomic differences in neonatal mortality and an increased inverse association between socioeconomic status and post-neonatal mortality.2 In the present study we estimate the association between parental education and cause-specific post-neonatal death and, in particular, secular trends in these effects.
|
Materials and methods |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Study population
This study was based on data from the Medical Birth Registry of Norway, which was established in 1967. Registration is based on compulsory notification of all live births, and stillbirths after 16 weeks of gestation and contains information on demographic items, complications during pregnancy and birth, as well as pregnancy outcome.3 Through linkage with Statistics Norway, the records are routinely updated with information on all deaths during the first year of life, including cause of death. These registries were linked to information on parental educational levels attained by December 31, 1998 by Statistics Norway.4
There was a total of 1 536 098 live births in Norway from 1969 to 1995. 52 241 infants with missing information on maternal and/or paternal education were excluded, yielding a study population of 1 483 857 live-born infants.2 Infants were allocated to three cohorts. The first cohort included 631 998 live births from 1969 through 1979 (the 1970s), the second cohort included 509 426 live births from 1980 through 1989 (the 1980s), and the third cohort included 342 433 live births from 1990 through 1995 (the 1990s).
Variables
The dependent variable was death in the post-neonatal period, i.e. from 28 to 364 days of life (4464 infants). Cause-of-death statistics was based on the underlying cause of death reported on the death certificate. The International Collaborative Effort on Perinatal and Infant Mortality (ICE) established a system for the categorization of causes of infant death into eight categories.5,6 Congenital conditions, infections and sudden infant death syndrome (SIDS) were the major causes of death (Table 1). In the present analysis, we combined the remaining five categories into ‘Other causes’ (Tables 2 and 3, and Figure 1).
|
|
|
|
Maternal age, parity, plurality, and parents' education were the independent variables. Mother's age was categorized as <20, 20–34, and >34 years. Parity and plurality were dichotomized; firstborn and later-born infants, single and multiple births.2 The highest exam passed, according to the International Standard Classification of Education (ISCED) was categorized into three educational levels in years (<10, 10–12, >12).2,4,7 For 110 305 (7.4%) births, information about the father was not reported in the Medical Birth Registry. These infants were included in the analysis and categorized in the variable ‘paternal education’ as ‘unknown fathers’ [Table 2, (b)].
Statistical analysis
All mortality rates were estimated as the number of deaths per 1000 live births according to the international standard. The association between parents' education and the cause-specific post-neonatal mortality was estimated as absolute risk differences and by multinomial logistic regression analysis.8 Live-born infants and the highest level of education served as reference groups. The covariates maternal age, parity, and plurality were included because they are established as factors influencing a variety of pregnancy outcomes. To control for the effects of the independent variables, adjusted odds ratios were estimated with 95% confidence intervals. Owing to the low mortality rates, the odds ratios can be interpreted as relative risks. Population attributable fraction (PAF) was estimated to assess the proportion of cause-specific post-neonatal deaths that would have been prevented if the risks among children of mothers with low and medium level of education were reduced to the risk among children of mothers with the highest level of education.9,10 The relative index of inequality (RII) was calculated as the OR from a logistic regression with the covariate given as the proportion with educational level higher or equal to one's own education level.10 This measure can be interpreted as the relative risk between the bottom and top of the educational hierarchy. SPSS statistical software was used in the analyses.8 The four measures represent aspects of social inequality with potentially different public health relevance and policy consequences.
To investigate variation over the study period we calculated mortality rates in each year of birth and for each level of maternal education. The rates were smoothed because cause-specific mortality was low and the relative variation was large. Smoothing was carried out in SPLUS with the routine for generalized additive models.9
|
Results |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
In the 1970s the major cause of death in the post-neonatal period was congenital conditions followed by SIDS and infections. In the 1980s and 1990s the major cause of death was SIDS. The largest reduction was observed for congenital conditions (risk reduction 0.57/1000) followed by infections (0.40/1000). Deaths caused by SIDS increased from the 1970s to the 1980s (0.86/1000 to 1.80/1000) and decreased to 0.72/1000 in the 1990s.
Table 2 shows the cause-specific mortality rates by maternal and paternal education. There was a reduction in absolute risk attributed to congenital conditions both among infants whose mothers had low education (
0.42/1000) and those whose mothers had high education (0.48/1000) in the period under study. The risk difference between infants whose mothers had low education and those whose mothers had high education was 0.25/1000 in the 1970s and 0.31/1000 in the 1990s.
There was a reduction in absolute risk attributed to infection among infants for all education groups. The highest rates were found in the group with unknown fathers, and the risk reduction was only 0.18/1000 from the 1970s to the 1990s. The risk difference between infants whose mothers had low education and those whose mothers had high education was 0.23/1000 in the 1970s and 0.34/1000 in the 1990s. There was a similar pattern for paternal education.
SIDS accounted for 27.2% of the post-neonatal mortality in the 1970s, 50.6% in the 1980s, and 37.7% in the 1990s. The absolute risk attributed to SIDS increased by 0.51/1000 among infants whose mothers had low education from the 1970s to the 1990s. In contrast, among infants whose mothers had high education, the absolute risk attributed to SIDS decreased by 0.56/1000. Infants with unreported fathers had high mortality rates, and an increased risk difference of 0.79/1000 from the 1970s to the 1990s. For SIDS, there was a reduction in all groups from the 1980s and a diminishing risk difference between education groups over this time.
For all causes of deaths, except for ‘SIDS’ and ‘Other causes’ in the 1970s, there was a persistent inverse association between maternal educational level and mortality, and the adjusted odds ratio reveals increasing inverse association over time.
The first row in Table 4 shows the risk for parents with low education relative to those with high education. For all causes of death, the risk for infants whose parents both had low education was highest in the 1990s relative to those whose parents both had high education (Row 1). The PAF confirms this pattern (Row 2). The proportion of deaths that could be attributed to low and medium maternal education increased from the 1970s to the 1990s. The pattern for RII was similar to that for PAF (Row 3).
|
Both absolute and relative measures of inequality reveal increased educational differences in cause-specific mortality from the 1970s to the 1990s.
Figure 1 shows the smoothed cause-specific mortality rates by year, according to maternal education. The mortality rates for congenital conditions and for infections declined steadily for all levels of maternal education, and in accordance with the risk differences reported in Table 4, the mortality rates were approximately parallel. However, mortality owing to infection among children whose mothers had high education had almost vanished by the end of the study period and this explains the high relative risk and PAF in Table 4 for infections in the 1990s.
In contrast, Figure 1 clearly displays the SIDS epidemic in the 1980s. The curves also show that there was very little difference in SIDS mortality in the early 1970s and that the epidemic in the 1980s was largely an epidemic among infants whose mothers had low-level or medium-level education. In addition, the curves indicate that SIDS mortality declined earlier for children of mothers with high education.
For the remaining causes, mortality differed minimally according to maternal education in the 1970s. Thereafter, the rates declined for infants whose mothers had medium or high education but remained at the same level or increased for those whose mothers had low education.
|
Discussion |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Our analysis demonstrates that the increased socioeconomic difference in risk of post-neonatal mortality is due primarily to an increase in SIDS among infants whose mothers and/or fathers had low education and infants whose fathers were not reported in the Medical Birth Registry. Social inequality widened for post-neonatal deaths caused by SIDS and infections from the 1970s to the 1990s. This was mainly due to a reduction in risk among those whose parents had medium and high levels of education. The socially disadvantaged groups constituted a smaller proportion of the population in the 1990s compared with the 1970s, and the risk of post-neonatal infant death was low in all social groups by any historical or international comparison.
An important question is whether the increased inequality we observe in the 1990s is a transient historic phenomenon. Will parents who have low education gradually achieve equally low death rates among their infants as those who have high education, or is it a persistent phenomenon that gets worse because parents with low education are now a smaller and more selected group than they were in the 1970s?
Methodological considerations
The size and completeness of the dataset is a major strength in this study.2–4 Strengths and weaknesses have been discussed in an earlier paper.2 In summary, it is unlikely that the results can be explained by selection bias, information bias, or changes in coding or classification in The Medical Birth Registry or the Education Registry.2–4
Secular variations in the coding of causes of death are likely. Norway converted from ICD-8 to ICD-9 in 1986 and from ICD-9 to ICD-10 in 1996. The study population covers births from 1969 to 1995, thus algorithms have been established for deaths coded by ICD-8 and ICD-9 (International Classification of Diseases, eighth and ninth revision). According to the ICE-system the causes of death in the preliminary analysis were categorized as; ‘Congenital disorder’, ‘Asphyxia-related disorder’, ‘Immaturity-related disorder’, ‘Infections’, ‘Other specific disorders’, ‘SIDS’, ‘External causes’, and ‘Remaining diagnoses’.5,6 The category ‘External causes’ includes all deaths caused by accidents, poisoning, violence, and nutritional deficiencies. The category ‘Other specific disorders’ includes specified conditions (e.g. neoplasms and endocrine disorders). The category ‘Remaining diagnoses’ encompasses deaths that were not allocated to any of the other categories (Table 1).5,6,11 Thus, the category ‘Other causes’ shown in Tables 2 and 3 is a very heterogeneous group and difficult to draw inferences from.
The autopsy rate increased in the period under study and may have affected the classification of death.12 However, it is well documented in Norway that the increase in SIDS from 1967 to the late 1980s was real and not due to changes in diagnostic practices.12,13 We have based the causes of deaths on the underlying cause of death reported on the death certificate, and changes in diagnostic practices cannot be tested.
Nearly all women undergo second trimester ultrasound in Norway. The decrease in deaths caused by congenital conditions might be due to widespread ultrasound foetal diagnosis followed by selective abortion. Norwegian legislation in relation to prenatal diagnostic testing is rather restrictive and there is unlikely to be a social gradient in the use of ultrasound. Moreover, official statistics on the number of abortions following prenatal diagnosis are sparse and incomplete, and no conclusions can be drawn based on our data.
Why has social inequality increased?
After the peak incidence of 2.69 per 1000 live births in 1988, the SIDS rate dropped to 0.6 in 1995.3 The sudden drop in the SIDS rate can be attributed to a reduction in the prevalence of prone sleeping. Prevalence of prone sleeping increased steadily in Norway from 7% in 1970 to 49% in 1988 and subsequently fell to 27% in 1990.14 A campaign against prone sleeping was launched in Norway in January 1990.15
‘The Nordic Epidemiological SIDS Study’, which was performed during 1992 through 1995, showed no difference in the proportion of infants sleeping prone according to maternal education.12 Another study showed that maternal age and birth order partly explained association between parental education and SIDS.16
The social gradient in smoking has increased over time because a smaller proportion of pregnant women with higher education smoke, while those with low education have maintained their smoking habits to a larger extent.17 We do not know how the increased inequality in smoking habits affects our findings. However, one may speculate that these trends in the socioeconomic patterns of smoking prevalence could be the most important explanation for the mortality differences revealed in this study, particularly when comparing the 1970s with the 1990s.18
Low education seems to be an important risk factor for SIDS, beyond its association with other risk factors as accounted for in the present study.
The decline in deaths caused by infections from the 1970s to the 1990s is most likely because of improved medical care, particularly for low-birth-weight infants, rather than declining ascertainment related to infections.11 Possibly SIDS cases were erroneously classified as infections in the 1970s since upper airway infections may play a role in the aetiology of SIDS.19
Infants with unreported fathers are probably those whose mothers are truly single. In the 1990s, they were probably not mixed up with cohabiting mothers or categorized with cohabiting mothers, while in the 1970s and 1980s they were most probably categorized with those who were not married but cohabiting. The social disadvantage of being a single mother may, therefore, be clearer in the 1990-group of mothers who have no recorded father for their child, while the actual social disadvantage in the different time periods of being a truly single mother is hard to evaluate based on these data.
|
Conclusions |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Increased social inequality for post-neonatal death is primarily due to increases in the absolute and relative risks of SIDS among infants whose mothers and/or fathers have low education and/or ‘truly’ single mothers. Social inequality has increased during the study period for SIDS and deaths caused by infections. It is not clear whether these results reflect stronger social selection or a ‘time lag’ between socioeconomic groups, which will eventually lead to equally low mortality in all groups if or when differences in smoking habits diminish.
KEY MESSAGES
|
|
References |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
1 Norwegian Statistical Yearbook. Oslo-Kongsvinger: Statistics Norway, 2002.
2 Arntzen A, Samuelsen SO, Bakketeig LS, Stoltenberg C. Socioeconomic status and risk of infant death. A population-based study of trends in Norway, 1967–1998. Int J Epidemiol 2004;33:279–88.
3 Births in Norway Through 30 Years (1967–1996). Bergen: Medical Birth Registry of Norway, 1997.
4 NOS C751. Norwegian Standard Classification of Education. Revised 2000. Oslo-Kongsvinger: Statistics Norway, 2000.
5 Center for Disease Control, National Center for Health Statistics. Proceedings of the International Collaborative Effort on Perinatal and Infant Mortality, Volume III. Hyattsville: Center for Diseace Control, Center for National Health Statistics, 1992.
6 Cole S, Hartford RB, Bergsjø P, McCarthy B. International collaborative effort (ICE) on birth weight, plurality, perinatal and infant mortality. Acta Obstet Gynecol Scand 1989;68:113–17.[Web of Science][Medline]
7 Kunst AE, Bos V, Mackenback JP. Guidelines on Monitoring Socio-economic Inequalities in Health in European Countries. Rotterdam: Erasmus University, 2001.
8 Norusis MJ. SPSS 10.0 Guide to Data Analysis. Upper Saddle River, NJ: Prentice Hall, 2000.
9 S-PLUS 4. Seattle, Washington: Mathsoft, 1997.
10 Hastie TJ, Tibshirani RJ. Generalized Additive Models. London: Chapman & Hall, 1990.
11 Daltveit AK, Vollset SE, Otterblad-Olausson P, Irgens LM. Infant mortality in Norway and Sweden 1975–88: a cause specific analysis of an increasing difference. Paediatr Perinatal Epidemiol 1997;11:214–27.[CrossRef][Web of Science][Medline]
12 Øyen N. Sudden infant death syndrome. Epidemiological studies in Norway, Sweden and Denmark 1967–1995. (Thesis) Norwegian J Epidemiol Suppl 1997:7.
13 Irgens LM, Skjærven R, Lie RT. Secular trends of sudden infant death syndrome and other causes of post perinatal mortality in Norwegian birth cohorts 1967–1984. Acta Paediatr Scand 1989;78:228–32.[Web of Science][Medline]
14 Irgens LM, Markestad T, Baste V, Schreuder P, Skjærven R, Øyen N. Sleeping position and sudden infant death syndrome in Norway 1967–1991. Arch Dis Child 1995;72:478–82.
15 Markestad T. Information about sudden infant death syndrome for the general public. Acta Paediatr 1993;389(suppl):124–25.
16 Daltveit AK, Irgens LM, Øyen N et al. Sociodemographic risk factors for sudden infant death syndrome: associations with other risk factors. The Nordic Epidemiological SIDS Study. Acta Paediatr 1998;87:284–90.[CrossRef][Web of Science][Medline]
17 Eriksson KM, Haug K, Salvesen KÅ et al. Smoking habits among pregnant women in Norway 1994–1995. Acta Obstet Gynecol Scand 1998;77:159–64.[CrossRef][Web of Science][Medline]
18 Nordtrøm ML, Cnattingius S, Haglund B. Social differences in Swedish infant mortality by cause of death, 1983 to 1986. Am J Public Health 1993;83:26–30.
19 Gilbert R, Rudd P, Berry PJ et al. Combined effect of infection and heavy wrapping on the risk of sudden unexpected infant death. Arch Dis Child 1992;67:171–77.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Brustad, T. Pettersen, M. Melhus, and E. Lund Mortality patterns in geographical areas with a high vs. low Sami population density in Arctic Norway Scand J Public Health, July 1, 2009; 37(5): 475 - 480. [Abstract] [PDF]
|
|
|
|
 |
|
 M. M. Morales-Suarez Varela, E. A. Nohr, A. Llopis-Gonzalez, A.-M. N. Andersen, and J. Olsen Socio-occupational status and congenital anomalies Eur J Public Health, April 1, 2009; 19(2): 161 - 167. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Arntzen, L. Mortensen, O. Schnor, S. Cnattingius, M. Gissler, and A.-M. N. Andersen Neonatal and postneonatal mortality by maternal education--a population-based study of trends in the Nordic countries, 1981-2000 Eur J Public Health, June 1, 2008; 18(3): 245 - 251. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. K. Singh and M. D. Kogan Persistent Socioeconomic Disparities in Infant, Neonatal, and Postneonatal Mortality Rates in the United States, 1969-2001 Pediatrics, April 1, 2007; 119(4): e928 - e939. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. D. SMITH Mensuration, Mendel, and a 19th century public health justification for US imperialism Int. J. Epidemiol., August 1, 2006; 35(4): 811 - 813. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||